http://www.informatik.uni-trier.de/~ley/db/conf/cvpr/cvpr2006-2.html
CVPR 2006 (vol 2)
http://doi.ieeecomputersociety.org/10.1109/CVPR.2006.191	79	Motion Patterns: High-Level Representation of Natural Video Sequences Duangmanee Putthividhya, Te-Won Lee Institute of Neural Computation University of California, San Diego 9500 Gilman Drive, La Jolla, CA 92093	acknowledgement activation adaptive adelson advances algorithm also america analysis anonymous applications asilomar authors based bases because behavior bell bergen better both bubbles cells cheng classification code comments comparison complete component components compression computation computational computer computers conf conference contexts correlation correspond cortex cvpr data decomposes decomposition denoising dependencies detection different dimensional discussions distribution dynamics edge eltoft emergence encode energy estimation excellent explore feature features fields figure filter filters first found framework from future give good hateren have helpful high higher hoyer hyun hyvarinen ideas ieee image images independent information into invariant jiucang karklin laplacian layer learning learns level lewicki like likelihood linear london making manuscript mapping maximum meaningfully mixture model modeling models motion natural network neural neurocomputing nongaussian nonlinear optical order pages park pattern patterns perception performs phase plan presented primary proc proceedings processing promising properties readable receptive recognition references representation research restoration results reviewers royal ruderman scenes scheme second seems segmentation sejnowski sequence sequences shift shinko show showing shrinkage signals similar simoncelli simple society some sparse spatio spatiotemporal statistical statistics structures subspaces such synthesis systems taesu task temporal term textures than thank that their these this topographic torbjorn types unifying using variance vayrynen very video videos vision visual which wiener work would yields
http://doi.ieeecomputersociety.org/10.1109/CVPR.2006.255	151	Robust AAM Fitting by Fusion of Images and Disparity Data	aams able acceptable achieve acknowledgements active adapt adding alignment also although analysis anger animation appearance applications apprehend approach approximate arya atively attributes author avoid aware away background baker based black blanz brary camera chai chang changes characteristic chil chosen cjlin classification classify closed color combined combining commission commun comparison computer conclude conclusions conditions conference consequence constraint context contract cootes could creation crossvalidation csie cvpr darrell data dealing deformations degenerate described despite detection developing different difficult disgust dominating during each ease easy eccv edwards emotion emotions enhanced estimation european even existing experiments expression expressions eyebrows face faces facial fear figure first fitting flow focus forchheimer form foundation frame frames from future generic given gordon grants greater ground harville have head holds holland however http icmi icpr ieee illustration images implementation improve improved improving incorporating increasing independent independently information inherently initial initialization instabilities instability interact into issues iter iterative john journal kalioubry kanade labeled learning least left library libsvm lighting line linearly loose lowresolution machines make manually matthews means meeting mesh meth method michel model models more morency moreover morphable motion mount movements moving mpeg muscle national nature nearest neighbor neutral noise normal notably numeric numerical obtained occlusion occlusions occurring order other over pages pairs pami pandzic parameter parameters partial partially particular pattern performed place points pose possible precise precision presence prevent previous proc proceedings processing project proposed qualitative quickly raising range ranges rate real realtime recognition recordings recovery references regression relaxes reliability remaining removal representations required requirements research resolutions result results revisited reweighted reweighting right rigid robust roughly same scaled scheme scholarship science searching separate sequence serve sets shape show shows siggraph significantly similar simple since smaller society solution some sons sorrow speed springer squares stability staged standard stat state statistical stereo still successful such support supported surprise surprisingly synthesis system table tackle taylor tenfold texture that them theor theory they this those three time tracking training transition truth turn unaware under unseen unsupervised usabil used using vapnik varying vector vectors vetter video vision well welsch were when which while white wiley with within woodfill work working world xiao yielding
http://doi.ieeecomputersociety.org/10.1109/CVPR.2006.114	90	Estimating Intrinsic Component Images using Non-Linear Regression Marshall F. Tappen Edward H. Adelson William T. Freeman MIT Computer Science and Artificial Intelligence Laboratory Cambridge, MA 02139 {mtappen, adelson, billf}@csail.mit.edu	academic adaptive additive adelson agreement america analysis anandan annals application approximation aston available barrow bayesian bell bellman bishop black boosting burt cambridge canada carmichael chang chapter characteristics cjlin collaboration computation computer conference control corporation csie cvpr deriving dynamic editors elad estimation evidence experts field framework freeman friedman from function gradient graphics greedy hanson hastie hinton horn http ieee image images index inference intelligence international intrinsic jacobs jordan journal keshet kimmel knill land learning level library libsvm lightness local logistic machine machines massachusetts mccann mixtures mosaics motion multiresolution nabney ncrg netlab network networks neural nippon nowlan optical over oxford pages part pasztor pattern pentland perception press princeton priors proceedings processes recognition recovering references reflectance regression retinex richards riseman robot robust roth scene sequences shading shaked single sobel society software spline statistical statistics support systems tappen telegraph telephone tenenbaum theory tibshirani time transactions university vancouver variational vector view vision volume weiss with york
http://doi.ieeecomputersociety.org/10.1109/CVPR.2006.244	132	ReconstructingOccludedSurfacesusingSyntheticApertures: Stereo,FocusandRobustMeasures Vaibhav Vaish Richard Szeliski C. L. Zitnick Sing Bing Kang	about acknowledgements acquisitions adams adelson after algorithms aloupis also analysis anandan andother antunez aperture apertures approach arrays artificial assistance author baker barth based baseline bayesian been begun behind belief believe bennett better beusefulforthereconstructionofnon blue bonet bosch bottom calibrate camera cameras case cause causing chai circles closer collins color coloring combined comparison compositing computed computer computing conference cost created cuts cvpr darpa defocus defocused deleted deleting dense depth develop different diffuse diss does during dyer dynamically each eccv effects eino enforced entropy estimation estimators existing experiment explains extending face factor factorization failure favaro favaroands field figure fill finally find first fitzgibbon focus focusing foreground four from functions garg geometric gift golland gortler grants graph graphics greg handle handlingocclusionsindense have high holes horowitz http iccv ieee ijcv image images imaging improve incorporate increases inproc inset internship interpolation into isaksen joshi kanade kang kiryati kolmogorov lambertiansurfaces large later layer layered layers left levoy light like limitation limits location makes manually marvelsofafinitelensaperture master matching matting mcgill mcmillan median mennucci microsoft model motion much multi multiple nbch number objects observing obtained occluded occluder occluders occlusion okutomi outliers pami papers parallax part pattern percentage performance person photorealistic pixels plane plants point points proc proceedings project projecting propagation quality really recognition recons reconstruct reconstructed reconstructing reconstruction redmond reference references regularization reparametrizedlightfields representation research responsibility rest reviewers right robust roxels scene schechner school science screening seeingbeyondocclusions seitz shape shear should showing shown shows siggraph size soatto society some somewhat space spatial stanford statue stereo strauss such suggestions supported surface surfaces sweep synthetic szeliski talvala texture than that their them theoretical these thesis they this threshold thresholds trans transform transparency true uniform university unoccluded using uswebsite uyttendaele vaish video view viewing views ville viola visible vision volume voxel voxels wall wang warp ways weighted wewouldliketothankaugustoroman wexler while wilburn winder winner wish with work workshop would zabih zisserman zitnick
http://doi.ieeecomputersociety.org/10.1109/CVPR.2006.264	111	Scalable Recognition with a Vocabulary Tree David NisterŽ and Henrik SteweniŽ us Center for Visualization and Virtual Environments Department of Computer Science, University of Kentucky	accepted across adapted affine also approach approximate association austrian based baseline bauckhage baumberg been berg bischof bmvc book both bottom bourne brightness building built camera captured case cdcovers changes chum classification clip colosseum comparison computational computer computing conclusions conference connected content contraints correspondence cover covers cues curse cvpr darrell database datar defined defining definition demonstration depth descriptors detectors different dimensionality ding directly discriminative displayed distinctive distortion distortions distributions easily eccv engine entropy estimation evaluation even example extremal faces feature features figure finds firewire frames fraundorfer from geometry google gool grauman ground hashing hierarchically hopeful however iccv identity ieee ijcv image images immorlica important improved including increasing independent indexing indyk interest internet invariant kadir kernel keypoint keypoints lagger laptop large larger lazebnik lead left lepetit less lindeberg linear local localitysensitive location lowe lucky make malik martin matas match matching maximally mikolajczyk million mirrokni modeling mohr more most motwani movies multi music nearest neighbors norm obdrzalek object occlusion pages pajdla pami pattern performance performed planar played point ponce powerful presented proc proceedings produced pyramid quantizes queries query quite randomized real realtime recognition references region regions reliable removing respect result results retrieval rich right rigid robust rotation rothganger running scale scenes schaffalitzky scheme schmid search searches searching second separate separated sets seven shape short shot shown significantly similarity sivic size smaller smoothing snapshot society some spatial specularities stable state stereo still structure subset successful such symp target text than that theory these this time timing towards tree trees truth tuytelaars upon using video videos view viewpoint views vision vocabulary volume ways weighting well were where wide widely with working works workshop zisserman
http://doi.ieeecomputersociety.org/10.1109/CVPR.2006.167	100	Learning Distance Metrics with Contextual Constraints for Image Retrieval Steven C. H. Hoi, Wei Liu, Michael R. Lyu and Wei-Ying Ma	adaptive algorithm allerton analysis annual application bach background based between bialek biased bottleneck cambridge cardie chang circuits civr classification classifiers clustering communication component components computer computing conf conference connection constrained constraints content contextdependent control cristianini cvpr data direct discriminant discriminants discriminative discussion distance early engineering enhancing equivalence exploiting face feedback fisher formulating from functions garg generative goldberger gupta hastie haussler hertz highdimensional hillel hinton huang icml ieee image improving independent information interactive international jaakkola jain jmlr john jordan journal kernel kernels knowledge learning lncs machine mahalanobis mclachlan means method methods metric mika models muller multidimensional multimedia murty nearest neighbor neighbourhood nips nonlinear novel pages pami panda pattern pereira press proc proceedings processing ratsch recognition references regression relevance research retrieval review rogers roweis russell salakhutdinov santini scaling scheme scholkopf schroedl shawe shental side signal similarity smeulders society spectral statistical surveys systems taylor technique technology theory tibshirani tishby trans unified university vapnik variants video vision wagstaff weinshall weiss weston wiley williams with workshop worring xing yang years york zhou
http://doi.ieeecomputersociety.org/10.1109/CVPR.2006.242	54	Recognition of Composite Human Activities through Context-Free Grammar based Representation M. S. Ryoo and J. K. Aggarwal Computer & Vision Research Center / Department of ECE University of Texas at Austin	action actions activities aggarwal allen analysis april articulated august based bayesian bobick body computation computer conference cviu cvpr detecting essa event events ferguson from hidden hierarchical hierarchy hongeng huang human ieee interacting interactions interactive interval issue ivanov journal language learning level logic markov media mining minnen models motion movement multimedia multiple network networks nevatia nguyen nonrigid ordered pami park parsing partially parts pattern persons phung pinhanez proceedings propagation recognition references representation review semantic sequential simultaneous society spaces special stochastic streams surveillance systems temporal thesis tracking trajectories transactions understanding using venkatesh video vision visual workshop zhao
http://doi.ieeecomputersociety.org/10.1109/CVPR.2006.323	53	Using Bilinear Models for View-invariant Action and Identity Recognition Fabio Cuzzolin UCLA Vision Lab, University of California at Los Angeles Boelter Hall, 90025 Los Angeles, CA	abdelkader acknowledgements acoustics acquah action activity adelson afgr afosr aggoun algorithm algorithms analysis analysisbased analyzing andlip annals approach approaches area automated automatic avbpa avsbs based basedrecognition bennett between bhanu bilinear biometric biometrics bobick body borer bowyer boyd carnegie carter cattin challenge characterization chellappa chowdhury classification collins colloquium combining comparison computation computer conference content control cunado cuntoor cutler cvpr darrell data database datasets davis debrunner dempster description eccv eigengait eigenspace elgammal elliot estimation evidence evidences evolution experimental experiments extraction face feature federal figures foster framework freeman frieze from frosini functions fusion gait gaitid gathering grimson gross grother halmstaadt harris hayfron hidden human humanid humans icip icpr identification identity ieee ijcv image improve incomplete individual information institute integrated integration intelligent international invariant johnson journal kale kinematicbased kruger kullback laird leibler lett likelihood little manifold markov math maximum measuring mellon methodfor metrics mobo model modelling models moments moore motion moving multi multiple multisensor murase nanda nash neural ning nixon niyogi nonlinear object other page pages pami parameters pattern patterns people performance performances performed periodic person phillips prgel problem proc proceedings processing rajagopalan ranging reading recogn recognising recognition recognizing references relationship report representation research robotic royal rubin running sakai sampling sarkar self sensor separating sequence sequences several shakhnarovich shape showing shutler signal silhouette similarity size society speech spie springer standard stat static statistical statistics style sufficiency sundaresan sunsaresan supported sweden swiss symmetry system systems tech technical technology tenenbaum test their tolliver tomatic towards tracking trans tsin university unknown urtasun used using vega velocity verlag view views vision visual volume walking wang when with workshop zlatnik
http://doi.ieeecomputersociety.org/10.1109/CVPR.2006.71	89	Blind Haze Separation Sarit Shwartz , Einav Namer and Yoav Y. Schechner Dept. of Electrical Engineering Technion - Israel Inst. Technology Haifa 32000, ISRAEL psarit@tx.technion.ac.il,{einav,yoav}@ee.technion.ac.il	acknowledgments adelson advanced alon amari amer analysis aperture application approach artificial asilomar atmosphere attenuation baker based bellingham bidirectional blind blinddehazing bmbf bofill bolas caimi center chen cichocki class classification clear comp component components compression computation computations computer computers conditions conducted conf conference confocal containing contrast convolutive cover cozman cula curila current cvpr dana decomposition degradation dept depth descatter design detection dictionary diffuse eccv efficient elect elements engineering entropy establish estimating estimation exchange extensions face farid fast fellow filtered foundation framework frequencies from funded garg glimpse godin grant graphics gross haze home horowitz hyvar ieee ikeuchi image images imaging improvement improving independent inen information instant inverse irani israel jaffe john journal karhunen karpel kernel kiryati kisilev kocak kopeika kroktov kumar landau layer layers learning levoy lewicki lighting lncs machine matting mcdowall minerva mixture mixtures miyazaki modalities model modelling models multilinear multiplexing multiscale namer narasimhan nayar needed neural noising nuzilland oakley objects oceanic ollendorff optimal optimization oriented other page pages pami past pattern pearlmutter photography physical poisson polarization poor proc proceedings processing pursuing quality rain raytracing recognition recovery references reflection reflections reflector removal representation representational representations research restoration roca sample sarel satherley scattering scenes schechner science segmentation semi sensory separating separation shamir shape shum shwartz signal simoncelli society sons source sparse specular spie statistical structure such supported surface synthesis synthetic system systems tang taub technion terzopoulos theory this thomas through torre training trans transparent treibitz umeyama underdetermined understanding underwater unsupervised using vaish vasilescu videos vijaya visibility vision volume wang wash wavelet wiley with wolff work worth yoav zeevi zibulevsky
http://doi.ieeecomputersociety.org/10.1109/CVPR.2006.198	114	Multi-Target Tracking ­ Linking Identities using Bayesian Network Inference Peter Nillius Josephine Sullivan Stefan Carlsson	algorithm applications approach approximate arising assignment association balch barros based bayes bayesian belief bmvc boosted bouthemy boyle cadre camera cameras carlsson class cohen computational computer computing conference congestion cowell cvpr darwiche data dawid dellaert detected detection doucet eccv european expert figueroa filter football formulation freitas from gelgon graph guide huang iccv icip icpr ieee image images inference integrating interacting international iwase jones journal kettnaker khan labelling lagrangian lauritzen leite little lowe maintaining matlab mcmc medioni mixture moving multi multidimensional multimodality multiple multisensor multitarget murphy needham networks objects occlusion okuma optimization orwell pages parallel particle pattern perez players pmht poore positions probabilistic problems procedural proceedings reasoning recognition recovery references relaxation representation robertson saito scale science sequence soccer society spiegelhalter sports springer statistics sullivan surveillance syst taleghani targets through toolbox tracking trajectories using vermaak view vision volume with zabih
http://doi.ieeecomputersociety.org/10.1109/CVPR.2006.160	42	Joint Recognition of Complex Events and Track Matching Michael T Chan Anthony Hoogs Rahul Bhotika Amitha Perera John Schmiederer Gianfranco Doretto GE Global Research One Research Circle Niskayuna, NY 12309, USA	abnormality account acknowledgments across acting action activities activity actor adaptive additional address addressed aggarwal algorithms also although analysis anomalous apparatus approach assignment assistance assume august background bags based batta bayesian behalf behaviour berkeley bobick both brand bremond brooksby busy california casee challenging chan changes characterize clutter coleman comaniciu combination complete completely complex computer conclusions conditions conducted conf conference constitutes contained contribution corporation correctly corresponding cviu cvpr damages demonstrate despite detecting detection disclosed discovery driver dynamic either enables essa estimating evaluation event events evidence example experimental experiments explanation expressed figure formulation frames framework from fuel gaps gong grams great green grimson group hakeem hamid have hierarchical hongeng hoogs huang hundreds iccv icpr ieee implementation implied indicate indicates inference information infringe integration intelligence interacting interactions intille involving isbell johnson joint june kaucic kaufhold kettnaker labelling layer learning liabilities limited linking links lockheed long machine makes manual martin mean meer method methods mining minnen mixture model modeling models more multiagent multiple munkres murphy natl neither networks nevatia november object objects occlusion occlusions optimizes order ordered overall owned pages paper park partially pattern patterns perera performance person petersen piece point potential preliminary prepared presence primary primitives privately probabilistic problem problems proc proceedings process profiling propagation property providing purple ramesh rare real reasonable recognition recognizing references report representation representing respect resulting results rights rigid robust robustness scene scenes schmiederer score segmentation semantic semantics sensor sequences sequential shah sheikh shift siam significant simultaneously sinks society solving some sources sponsored stauffer technical thank that these thesis this thousands through time together track tracking tracklets tracks trans transportation truck underway unified university using validation very video viewpoint violet vision visual volume want warranty wensheng which with without work workshop world xiang zhaohui zhou
http://doi.ieeecomputersociety.org/10.1109/CVPR.2006.188	34	Modeling and Classifying Breast Tissue Density in Mammograms Anna Bosch, Xavier Munoz, Arnau Oliver and Joan MartiŽ ~ Computer Vision and Robotics Group University of Girona Campus Montilivi, Girona {aboschr,xmunoz,aoliver,joanm}@eia.udg.es	abnormalities according acknowledgements affect affine aided american analysis andrew appearance approach aspects astley automated automatic avoid background banks beijing berlin besides betal better biology birads blot boggis bosch bovis bowyer boyd brady breas breast buseman byers byng calonge cancer cancun carcinoma categories category cerneaz chan china chinga classification classifier classifying clustering college combined computed computer computerized conference cvpr dance data database dataset densities density denton descriptor descriptors detection detectors determined development different digital dimensional discovering discussions distinctive distribution domain eccv edition efros elsa embs estimation estoril extraction facts features ferrari figures filter fishell founded freeman freixenet from further gatica germany girona goodsitt gool grant hadjiiski have heath helvie hence hofmann hospital hutt ibpria iccv icip identification ieee ijcv illustrated image images imaging inclusion increase indexing influence information interest international invariant iwdm jong josep june kadir karssemeijer kegelmeyer keypoints kopans labeling latent learning lesions leung local locations lowe machine madison malik mammogram mammograms mammographic mammography mart masses materials matters medical medicine mexico mias microcalcifications mikolajczyk miller miua model modeling monay moore moreover mouchawar munoz muscle necessary negatively objects october odobez oliver optimal over overlap pages paradigm parameters parenchymal parker partially pattern patterns pectoral perez petrick petroudi physics plsa point pont portugal present previous probabilistic proceedings quelhas raba radiology rangayyan recognition recognizing references remaining reporting represented representing research results retrieval ricketts risk russell sahiner same savage scale scene scenes schmid screening segmentation semantic shown sift sigir similar singh sivic society spicular stamatakis standard statistical studied study suckling surgery system taylor techniques textons texture than thanks that their these this three tissue together trueta tuytelaars university unsupervised using various varma very vision visual volume want when which wisconsin with wolfe women work working works yaffe young zhou zisserman zwiggelaar
http://doi.ieeecomputersociety.org/10.1109/CVPR.2006.159	16	Joint Boosting Feature Selection for Robust Face Recognition Rong Xiao1 Wujun Li2 Yuandong Tian3 Xiaoou Tang1	acoustics adaboost additive algorithms america analysis annals annual another audio authentication automatic based bayesian beveridge biometric blanz bolme boosted boosting bunch cascade chain chellappa chen christoph classification classifier cognitive complete complicated compression computational computer conditions conf conference confidence cvpr database dataset daugman detection direct directions discrete discriminant draper dual effect efficient eigenfaces elastic enhanced etemad evaluation expression extended face faces facial feature features feret fisher focus freeman friedman future gabor gender gesture gestures graph hastie huang human identification ieee illumination image images improved integrate intelligence international jean jensen jiang jones journal kittler kullback laurenz learning leibler lighting linear logistic luettin machine maitre marc matas matching measure messer methodology model modeling moghaddam moon more morphable multiclass murphy nastar networks neural neuroscience norbert object optical pages pattern pentland person phillips predictions proc procedures proceedings processing promising purpose rapid rated realistic recognition reconstruction references regression sample schapire scheme shan shannon sharing shows shum siggraph signal similarity simple singer society space speech statistical statistics structure synthesis synthesize system systems tang teixeira theory tibshirani torralba trans transform turk unbalanced used using vetter video view viola virtual vision volume vtsdb wang wechsler which will with work workshop yang zhang
http://doi.ieeecomputersociety.org/10.1109/CVPR.2006.293	137	Stereo Matching with Symmetric Cost Functions Kuk-Jin Yoon and In So Kweon Robotics and Computer Vision Lab. Dept. Electrical Engineering and Computer Science, KAIST 373-1, Guseong-dong, Yuseong-gu, Daejeon, 305-701, Korea	according account acknowledgments adaptive aiming algorithm algorithms although analysis applied applying approach approximate ation based because belief birchfield bobick border both boykov bpbased chai color compact comparable compared comparison computation computer computing conclusion conference conventional cooperative correspondence cost cuts cvpr cycle data defining dense dependent design detection diffusion disc disparity display dissimilarity energy error evaluation even experiment experimental experiments explicitly fast figure finally foundation frame freeman from function functions global good gradient grant graph greatly ground half handling have highly however identical ieee ignore image images improve improved improvement improving industrial insensitive integral intelligence intensity international intille into journal kanade kang kolmogorov korea korean kweon laboratory large left likelihood machine maps marr matching measure method methods middlebury minimization minimum ministry model modeling multi national nonlinear nonocc number occluded occlusion occlusions okutomi only optimal other pages paper parameters pattern performance pixel pixels poggio prior proceedings program propagation proposed ratio really recognition reference references research result results sampling sawtooth scharstein science search segm segmentation show shown shum society standard state statistics stereo support supported symm symmetric szeliski table tappen target taxonomy technology testbed textured than that then theory this tomasi took transactions truth tsukuba untex using variable veksler venus verified view vision visual volume weight well when while window windows with without worse yoon zabih zheng zitnick
http://doi.ieeecomputersociety.org/10.1109/CVPR.2006.22	28	A detection technique for degraded face images Shinji Hayashi1 and Osamu Hasegawa1	adaboost ahuja although among applied around baker baluja based bernt better blake bmvc boosted boosting both bottom calculated cascade cascaded cbcl centric computer conclusions conference context conventional correctly cvpr data december detect detected detecting detection detector detectors detects does eccv effective efficient engenders equals evaluation expand expanded expanding extracted face faces false fast feature features figure fourth frames from function future goal hallucinating hannes huang icafgr iccv identical ieee illumination image images impoverished improve increasing infer input invariant involves january jones july june kanade karen kriegman kruppa kullback learning leibler local lowresolution many means memo method might movie multi network neural number object only osadchy pami pattern pedestrians person pixel portray pose positives proc proceedings process processing produce proposed rapid rate rather real reason recognition references remains result results rotation rowley same scanning schiele schneiderman shina shown shum simple society statistical study survey suspicious technique that this those thresh time torralba trans under using varying view viola vision warning were when with work world yang zhang
http://doi.ieeecomputersociety.org/10.1109/CVPR.2006.299	135	Surface Geometric Constraints for Stereo in Belief Propagation Gang Li and Steven W. Zucker Department of Computer Science Yale University	about academic achieve acknowledgments active afosr afrl algorithm algorithms aloimonos also analysis anonymous approach approximate based because belief better birchfi blake both boykov buhmann cambridge camera carmichael carmo cient cipolla combination comments comparison compatibilites compatibilities computer computing conclusion conference consistency constraints contextual contour contourbased correspon correspondence critical curved curves cuts cvpr daniel darpa deformed demonstrate dence dense depth derive derived detailed detected devernay differential dimensional disparities disparity dissimilarity early eccv edition effi elementary emmcvpr employed energy error evaluation evidence example experimental explanations explicitly extend extends fast faugeras feature felzenszwalb field finally first flannery flow formulated foundations frame freeman frohlinghaus from generic geometric geometry giblin graph hall handle hartley helpful horizontal hummel huttenlocher iccv icpr identical ieee ijcv illustrate image images implications importance improved incorparating inference insensitive introduce iteration kass koenderink kolmogorov labeling layered learning level lncs local markov marshall matching measure message minimization model modeling models motion multi multiple multiway neill normal note numerical occlusion occlusions ogale order organization packages pages pair pami paper parameters particular pasztor pattern perceptual perform phasebased pixel position powerful prentice press problem proc proceedings processes product propagation properties proposed random reciples recognition reconstruction references regularizing relaxation research result results reviewers richard running sampling scene scharstein score second seconds shahar shape shapes shum similarly slant slanted snakes society software solid solution solved springer statistics stereo stereoscopic such supported surface surfaces szeliski tappen taxonomy terzopoulos teukolsky texture thank that their this three thresholds time tomasi traditional trans types univ university updating using veksler verlag vetterling view vision visual well with without witkin zabih zheng zisserman zucker
http://doi.ieeecomputersociety.org/10.1109/CVPR.2006.268	153	Seamless Image Stitching of Scenes with Large Motions and Exposure Differences	after agarwala aggarwal agrawala ahuja america amount anal annual apparatus approximate areas artifacts august beijing being better bilateral blake boykov bracket bracketed brown calibration camera cameras caused channels chen china cohen colburn color combined combining compensation composite compression computer cone conference contains could covered cream curless cuts cvpr data davis debevec december despite device differently digital display domain dontcheva dorsey drucker durand dynamic eccv eden editing eighth electronic eliminating energy especially european exposed exposure exposures extend extended extending fast fattal february field figure filtering final first framework from functions further gain gangnet generalized ghosting given goldman good gradient graph graphics grayish grossberg hand happened hara have high iccv ieee ijcv ikeda image images imaging information input inputs intell interactive international japanese jojic july june kanade kang kenbo kimura kolmogorov label labels lack large less levin lischinski litvinov lowe mach malik mann maps march minimization minimizedvia mitsunaga models mosaicing mosaicking mosaics most motion moving much nayar near ninth notice objects october only optical order outside pages panorama panoramas panoramic pass patent pattern peleg perez photographic photographs photomontage picard pickup pictures pixel pixels poisson prague probability proceedings process processing produced radiance radiometric ramesh range recognising recognition recovering reference references region reinhard reproduction response result results right saito salesin saturated scene scenes schechner seamless second self sensor september sequence series shiba side signals society space spatially springer statistical stitching szeliski technique tenth than that there this those tone tonemapped tones trans transactions triplet tsin undigital uyttendaele values variable varying veksler verlag video view vignette vision volume warped weiss werman what which wide winder window with zabih zomet
http://doi.ieeecomputersociety.org/10.1109/CVPR.2006.138	49	Human Carrying Status in Visual Surveillance Dacheng Tao1, Xuelong Li1, Xindong Wu2, and Stephen J. Maybank1	abdelkader academic achieve achieved activity algebra algorithms also america analysis angle application automatic averaged backpack based baseline best bhanu biological bobick body bold bombay bone boston both bowyer boyd bross cannot carrying carter cells challenge chellappa chopra chowdhury classification clear codes collins comparing comparison computer conclusions conf conference content contrast converge convergence converges correcting corresponds cortical cunado cuntoor curves cutler cviu cvpr dalley data daugman davis december decomposing derau description detection developed different dimensional discriminant discriminatively dlda drought during effective ensembles error estimation evaluations examination experiments extraction face feature field figure filter finding first fractional framework france freeman from fukunaga fully fusion gabor gait gaits gallery general ghaderi give gives gordon grimson grother gtda hadsell haritaoglu harwood height huge human humanid humans iccv icpr identification ieee image important improved improvement india inferred information introduced introduction issue iteration iterations janardan johansson johnson joint journal kale katholike kauai kernel kittler kory kruger lathauwer learning lecun leuven linear little machine madison mainly manners marcelja matas mathematical maybank means methods metric mining model models motion movement multi multilinear multiresolution murray nice ning nips nixon normal normalized november numbers objects only optical orientation oscillations output paper parameters part pattern patterns pedestrian people perception performance performances periodic periodicity person phillips picard plane platform polynomial posed power press previous probe problem proc procedure proceedings processing profile proposed proposes psychophysics purposes quar quebec rajagopalan rank rate receptive recognition recognizing reduce references refinement reported reports representation representations required research resolution respectively responses results robust rorr rows same sarkar scheme schemes seen separating sequences sets shape show shown signal significantly silhouette silhouettes similarity simple society solve space specific spectral stage state static statistical status stride study style submitted subspace sundaresan supervised surgery surveillance symmetry synchronization synthesis table tables tanawongsuwan tasks templates temporal tenenbaum tensor terms terzopoulos that their these thesis this three through tieu time tpami traditional training trajectories trans troje uarq universiteit upon using usually vasilescu vega veres verification videre vision visual walking wang washington well what which while wideatt with
http://doi.ieeecomputersociety.org/10.1109/CVPR.2006.13	13	3D Face Recognition Using 3D Alignment for PCA	adaptive algebra algorithm analysis annotated applications approach april arms august austin authentication based basso belhumeur besl beveridge biometrics blackburn blanz bolme bone bowyer cambridge challenge chang chellappa circuits class classes colbry colostate component computer computing conference cvpr data december deformable diego dimensional draper eigenfaces eigensurface erlebacher eurographics evalfacerec evaluation example experiments expression expressions face faces facial fall february figure fisherfaces fitting flynn fraction frgc from frvt granada grand grother guide handling hausdorff heseltine hesher hespanha hoffman http icip icpr identification ieee image images imaging jain july june kakadiaris koch koudelka kriegman learning linear literature little march marques mckay method micheals modal model morphable multi multimodal murtuza neutral novel object october overview pages pami passalis pattern pears phillips poggio prescreener presence principal proc proceedings processing projection radial range reanimating recognition references region registration report rigid rosenfeld russ scruggs selection september shapes signal single society spain specific srivastava strang survey surveys symmetry symposium synthesis system systems tabassi tallahassee technique teixeira test theoharis thomas three toderici user using variation vendor version vetter video vision volume worek workshop yahagi yuan zhao
http://doi.ieeecomputersociety.org/10.1109/CVPR.2006.117	15	Face Recognition using 2.5D Shape Information William A. P. Smith and Edwin R. Hancock Department of Computer Science, The University of York {wsmith,erh}@cs.york.ac.uk	acquiring algorithm algorithms analysis appearance arbitrary based basic basis basri belhumeur blanz bronstein bruce building case chellappa chichester closeness closeto combines complement computer conclude cone conf conference consistency constraints converse courtesy cvpr data database dovgard eccv effects efficient eigenfaces enforcing existing exists experimental explored extreme face faces facial field first fitting fletcher florida form francisco frankot freeman from fromshading frontal geodesic geodesics geometric georghiades hancock harmonic have having hill however huber human humanid ieee illumination image images imaging infer information insensitive integrability international invariant irradiance jacobs joshi journal kimmel kriegman lambertian lighting linear manifolds many marr measurements method methods model models morphable needle nonlinear normals obtained only pages pami particular pattern pennec pentland perception performance pizer poor pose presented princi principal probabilities probe proc proceedings processing provides psychology rather realistic recognition recovered recovery references reflectance results riemannian robust route samaras sarkar shading shape signal significantly simpler single society south spherical statistical statistics structure study subspaces sudeep suggest surface surfaces symmetric synthesis tampa than that these threedimensional tools training trans turk under university unknown used using variable vetter vision volume wang when where which wiley workshop worthington zhang zhao
http://doi.ieeecomputersociety.org/10.1109/CVPR.2006.152	103	Incorporating the Boltzmann Prior in Object Detection Using SVM Margarita Osadchy and Daniel Keren Computer Science Department University of Haifa Mount Carmel, Haifa 31905, Israel (rita,dkeren)@cs.haifa.ac.il	adaboost advances advantage agarwal algorithm also although american anal analysis annals anti antifaces application applied approach architecture artificial association asymmetric auer average background bakir baluja based bayesian because bengio bensmail biological boltzmann boosting both bouveyron breiman burges calculations cambridge cars cascade case categories celeux center choice chooses cjlin class classification classifier classifiers clearly clustering computational compute computed computer conference constrained corresponds csie curve cvpr data database decomposition deficient delakis delalleau detection detector dimensional direct discriminant discriminative distribution duda during each edition efficient eigenfunctions eigenvalue energy equal error estimation european event example examples experiment exploiting extending face faces fast faster feature features fergus figure fink form frame framework franz freund from function fung further fussenegger garcia gaussian geman generative generic gibbs girard girosi good gotsman hart haussler helps heuristics high higher however http human hypotheses iapr icann ieee image images importance improve incorporating information informatique instability intell intelligence international interscience into invariances irani jaakkola jones journal kanade keren kernel kienzle knowledgebased laboratory larger learning lecture lecun lesser levi libsvm like linear mach machine machines mangasarian manor martin mathematics means method methods metrics miller model models montre much mullin must nature networkbased networks neural nips nonlinearly notes novel number numerical object opelt optimization order osadchy osuna ouimet outperformed outperforms overfitting pages paiement paradigm partement pattern performance perona pinz plan poggio point pontil pose powell press prior probably proc proceedings processing protocol pseudo quantify rank rare rate rationnelle recherche recognition reduce references refining regularized rehg relaxation relevance report representation restoration results robust roth roux rowley runtime same schmid schneiderman scholkopf science selection sequences seventh shavlik shot show shows significantly single small smaller society sparse spectral springer stabilization stacking statistical statistics stochastic stork suggested summarizes summary sung support surprising synergistic systems table technical test than that them theory thesis this three through thus trained training trans transactions trial trials universite unsupervised uses using usually utilizing vapnik variability variances vector vectors verlag verri video viewbased vincent viola vision volume watson weak weiss when wiley with wolf york zelnik
http://doi.ieeecomputersociety.org/10.1109/CVPR.2006.295	122	Structure and View Estimation for Tomographic Reconstruction: A Bayesian Approach	absolute acad acknowledgments addition address addressed after agarwal alfred algebraic alignment allows already america anal analysis angular applicability applications approach arbitrar arbitrarily arranged arun aspect assemblies assignment assymetric asymmetricobjects atomic australian authors automated automatic available baldwin bandwidth based basis bayesian belongie bibby bolstein borg bracewell career center chiu chuang circular closed closedform column columns comaniciu common comparison computer computing conducted conference congr contract contrast converge correlation cryo cvpr data dataset david decades denote department derivation determination determining different dimensional directions dkent done dorin elec electron eman energy entire error estimates estimating estimator farrow feature fellowship fields figure finance first fisher fitting form formulation fourth frank from future gelfand goncharov grant groel groenen gross ground have health heel help high higham hilden hope horn hours huang identical ieee images improves ingwer initial institutes integration intell intelligence interesting intern issues iteration jakana josa journal kernel known kriegman lauren least like lines llth ludtke mach machine macrmolecules macromolecular mallick mardia marin matching matrices matrix mean meer method microscopy mirco model modern molecular moments more much multidimensional multivariate mutual nandhakumar national nearest negahdaripour nicholas noisy novel nummerical objects obtained october optical optimal orientation orientations oriented orthonormal ottensmeyer oxford package pages pami paper parameters part particle particles particular patrik pattern penczek peter point posteriori practical presented press prior priors problem proc procedure proceedings processing program projection projections providing quart quaternions question radermacher radioastronomy radon random randomly recognition reconstitution reconstruction reconstructions references refinement related relative research resolution resource resources result robust routines sameer satya scaling scott second semiautomated serge sets shape shift show shown shows side simpler simultaneously single sloan society software solution song sophisticated space spatial spider springer squares stages stagg starting starts statistical strip struc struct structure supported surface than thank that their theoretical theory these this three through tomography tomorgraphic tool toward towards trans transform truth ultramicroscopy uncalibrated under unit univ university unknown used using vainshtein verlag view viewing views visible vision visualization were which while widely with work would
http://doi.ieeecomputersociety.org/10.1109/CVPR.2006.222	36	Performance Modeling and Prediction of Face Recognition Systems	acknowledgement actual affect alex algorithms analysis analyze apply automatic baback based beveridge bhanu biometric bobick bolme boult bowyer bruce challenge chang charles chellappa ciomputing classification colorado comparison computational computer concept conclusion conf conference corinna cortes craig cukic cvpr data david department described design detection draper duda elham estimate evaluation experiments face faces failure features feret fingerprint flynn from future galleries gallery gestures givens grand grant green grother group guide hart hoffman homeland human hyeonjoon icassp iccv identification ieee image improve individual institute intelligence international intl introduce iris jain james john johnson ketkar larger learning level literature machine maio maltoni marcio marques matching matthew measurement methodology methods model modeling models moghaddam moon national networks nistir object other output overview paper part pattern peng pentland perfect performance personal philips phillips polytechnic predict predicting prediction present presented probabilistic proceedings provide publisher qiang quality query rauss recognition references rensselaer report representation research results richard rizvi rong rosenfeld ross safety schmid science score scores scruggs second security similarity singh small society sons springer standards state states statistical stork support supported survery survey system systems tabassi task technical technology teixeira this three training trans tswg united university users using validation vapnik various vector version vision visual vladimir wang watson wayman ways weiliang will willey wilson without worek work working workshop xiang zhao
http://doi.ieeecomputersociety.org/10.1109/CVPR.2006.143	81	Image Denoising with Shrinkage and Redundant Representations	academic adaptation adaptive adelson advances algorithm algorithms analysis appear application applied april asilomar astronomical astronomy astrophysics atomic atoms august based basis bayesian berlin biometrika block both brox bruce candes chambolle chang chen chosen coding coefficients communications complexity component compression comput computer computing conference constraint constructing context contourlet contourlets coordinate cope coring curvelet cvpr data daubechies december decomposition defining defrise denoising devore differences diffusion directional discussed disparate domain donoho efficient elad electron embedded equivalence figueiredo focuss framing france frequency from future gaussian gaussians generalized gill gorodnitsky graph grove guleryuz have here ideal ieee image imaging implementation implications information international invariant inverse iterative johnstone journal june laussanne lecture limited line linear localization lucier lvii matalon mathematics methods minimization mixtures modeling morphological moulin mrazek mrzek multiresolution multiscale murray necessity noise noising nonlinear nonparametric norm norms notes november nowak numerical only optimization ours overcomplete pacific pattern physics play portilla practical press priors problem problems proceedings process processing pure pursuit pyramids recognition reconstruction redundant references regularization relaxation removal rennes representation restoration reweighted role rotationally same sardy saunders scale schemes science scientific search separation september shapiro shrinkage siam sides signal signals simoncelli society soft sparse sparsity spatial spatially springer starck stat steidl strela subsequent switzerland systems their theory these thoroughly thresholding thresholds through toolbox total trans transaction transactions transform transforms tseng using variation variational very vetterli vision wainwright wavelet weickert weighted welk will with work wright york zerotrees zibulevsky
http://doi.ieeecomputersociety.org/10.1109/CVPR.2006.218	52	Parameterized duration modeling for Switching linear dynamic systems Sang Min Oh James M. Rehg Frank Dellaert GVU Center, College of Computing, Georgia Institute of Technology {sangmin, rehg, dellaert}@cc.gatech.edu	aaai acknowledgment acoust algorithms american anal annual approach artech artificial association audio austin australian automatic automatica award awarded balch banff bayesian biswas blackwellised blackwellized blake boston bregler canada career cham character chiuso christoph classification cluttered comments complex computer conf conference control cvpr data decision dellaert detection diagnosis digalakis doucet driven dynamic dynamical dynamics econometrics efros eigentracking environment estimation fault figure filter filters frank from gales gibbs gordon graphics helpful heskes hierarchical hodgins house howard human hybrid icassp iccv identifiability ieee impact inference information intell intelligence intelligent intl isard issue james jebara journal jump khan kimball koller krishnamurthy learned learning lerner limits linear london machine markov model modeling models motion murphy natural network networks north observability october ostendorf pages parametric parr part particle patrick pattern pavlovi practical principles probabilistic proc proceedings processing quantifying recognition recognizing references rehg reviewers rosti rottscher russell sampling samsung sang scholarship seattle segment sequences shalom shum shumway siggraph signal soatto society software special speech state statistical stochastic stoffer supported switching synthesis systems techniques texture thank their theoretical this timeseries tracking trans transactions trees twolevel uncertainty unified using vidal video view vision visualization volume wang with work zoeter zweig
http://doi.ieeecomputersociety.org/10.1109/CVPR.2006.171	99	Learning Object Shape: From Drawings to Images	active addison analysis anguelov animation application applications approach autoclass automatica bayes bayesian belief belongie berg best binford canny categories cheeseman class classification classifiers collins comparison completion computational computer conditional conference contour cooper cootes correspondence cortex cortical coughlan cover cviu cvpr darrell data databases daugman davis deformable degroot description detection dimensional discriminative distortion distributions dleung eccv edge efficient elements external features felzenszwalb fergus ferreira fields filters finding first flow force freeman frequency generative gradient graham grauman guibas hidden huttenlocher iccv icsi ieee ijcv image induction information inspired intelligent invariant john jordan josa kaufmann kelly kernels koller kumar learning logistic loopy malik markov match matching merging metric model modeling models morgan naive nips object omohundro optimized orientation pami pattern pearl people perception perona pictorial poggio prince probabilistic probability proceedings propagation pyramid quattoni random reading reasoning recognition references regression relations report resolution rissanen rodgers rubner scale scape segmentation self serre sets shape shapes shortest shot siggraph snakes society sons space spatial srinivasan statistics stolcke structures stutz system systems taylor technical texture their theory thomas thrun tomasi torr training uncertainty unsupervised using vector vision visual wesley wiley with wolf york zisserman
http://doi.ieeecomputersociety.org/10.1109/CVPR.2006.155	73	Instant 3Descatter Tali Treibitz and Yoav Y. Schechner Dept. of Electrical Engineering Technion - Israel Inst. Technology Haifa 32000, ISRAEL ttali@tx.technion.ac.il , yoav@ee.technion.ac.il	about above academic achieved acknowledgements acquired active addison addition adelson advanced after alfano alon also amer amphoto amplification analysis analytic angles aperture appendix applying approach aquavideo assumptions attenuating avoid backscatter backscattered backscattering based battery beam become before being below between beyond bidirectional biologic biological biology biomed blind bmbf bolas bonnier books bottom build bulbs caimi calibration camera catalog causing center chang chapter cheap chen choices choosing circular clear clusters coherence combined comparative compare compensation components computer concavities conf conference confocal conjunction considerations consists construct containing contrast cronin cula cummings current currently cvpr damaschini dana degree demos depolarization dept depth descatter design designed destroyed detailed detection deviation diego difficult diffuser diffusers diffusing discrimination discussion discussions distance distant dives done dori easy edition efficiency einav either elect electron emission emit enables engheta engineering enhanced enhancement enough epifanie estimate estimated estimating even exists exit expensive exper experimental experimentally experiments exploration expression extends facing falloff farid fast fellow field figure filtered filtering filters fitting flash flicker flitton fluctuations fluorescent forand former foundation fournier from fruitful fundamental funded further gated gating general generate giakos gilbert glass glimpse grant great grey ground haim hall hamamatsu hand harsdorf have having haze hecht help hertzberg high higher highly holographic hopcraft horowitz hour housing iccv ieee ikeuchi illumination image images imaging implications improvement improving incandescent including independent index instant instr intensity inverse israeli jaffe jakeman jarry jordan journal jurczak kaiser karpel kattawar kermany kiryati kocak kokhanovsky koppal lamphead lamps landau large laser lasers last lasts latter least left less letters level levoy lewis light lighting limit linear linearly long loss lower mackintosh maitland make matched matches matrix mcdowall mcglamery measurement media medium memory method methods might minerva miyazaki mobley model modelling more mostly mounted mueller multiplexing multiply nadav namer narasimhan narrow natural nayar nearly newport noise noisy nondiffusing note object objects oceanic oceans often ollendorff omer once optical optically optics optimal oriel original other overruled pace page pages partial parts passive past pattern patterns performance pernicka phase photnics photography phys physics pine placed polarimeter polarimetric polarization polarizationdifference polarized polarizer polarizers power practical press proc proceedings projection projectors projects prone propagating propagation pugh quality radiative rakovic ramamoorthi range ranges rapidly rather raytracing reaching real realtime recharging recognition reconstructed reconstruction recovery reduction references reflections reflector refractive relative remote removal rendering requirements requires research resources results reuter right roberts robinson rough rowe sabbah sandblusted sankaran saturates scan scanners scattered scattering scene scenes schechner science scramble sealed seawater semi sensing separating separation several shallow shamir shape sharp shashar short shown shows shrimp shwartz signaling signature simulated single skerry smaller society source sources spaced spatial spectra spie spread springer stability standard started statistical steimer still stimulated strand structure structured study submersible successful sufficient supernova supported surface swartz synthetic system systems target targets taub taylor tech technion technique temperature temporal term testing tests than thank that their theoretically there this through throughout thus time tissues tlsx toneb transfer transmission transparent true tube turbid uncontrolled underwater uniform used using vaish varies verified very virtues visibility vision volume walker walsh wang water waters watertight weitz wesley when where which while wideband widefield widen wiley windows with without wolff work worth xenon yelin yield yoav york
http://doi.ieeecomputersociety.org/10.1109/CVPR.2006.311	39	Towards Multi-View Object Class Detection	ability able according accumulated acknowledge acknowledgements across activation adaptive affine affinely also analysis appear applied approach auer authors autom bags bank bart based baseline besides between boosting both bray british byvatov cars case categorization causing challenge challenges changes chum class classes clustering codebook codebooks composite computer conclusion conditions conf conference constraints contrary contrast contributions correct corresponding counterparts crowded csurka cvpr dagm dance descriptors detection detections detector detectors difference different dorko each eccv efficient einhaeuser estimation european evaluation everingham evidence exactly example experiment experiments exploration extend extremal face faces fall false fast feature features fergus ferrari figure finally flanders floatboost fragments france freeman from further fussenegger future generic gesture gestures gong gool gratefully grenoble have heads however human hypotheses iccv ieee image images implicit includes increase increases indeed independent information instances integrate integrates integrating interest international interpolate introduction invariant kanade keypoints large lazebnik learning leibe links local lowe machine matas matches maximally mean mechanism method mikolajczyk missed model modeling more motorbikes multi multiclass multiple multiview murphy naturally need nition novel number object only opelt ordered other outperforms output over pajdla pami parts pascal patches pattern pedestrian performance perona pinz point ponce poor pose poses positives proc procedures proceedings project properly proposed real recog recognition reduced references regions relaxes remove robust robustness rothganger same scale scatter scenes schiele schmid schneiderman search seemann segmentation selected selecting selection shape sharing shift shoes show significantly similar simultaneous single slight society spatial sphere stable stage state statistical stereo such support suppressing system systems taken test theart their three time torralba tracking training transfer transfering transferring tuytelaars ullman unsupervised urban using variations vary vector verification view viewpoint views vision visual vote votes weak weber welling when wide will willamowski with within work working workshop would yield zhang zisserman
http://doi.ieeecomputersociety.org/10.1109/CVPR.2006.128	92	Graph Laplacian Kernels for Object Classification from a Single Example Hong Chang & Dit-Yan Yeung Department of Computer Science, Hong Kong University of Science and Technology	adjacency advances allinson also american analysis annals applicable application applications apply arbitrary asymptotics based basis between bonn both bousquet bruce brunelli buhmann carlo chang chapelle characterizing chen chung class classification cluster clustering compute computer concluding conference conjectures connectivity considered consistency constructed cvpr data database editors effective effectiveness eigensignatures energy example experimental experiments expression extension face faces features fink fischer flda fogelman from future general germany ghahramani global graham graph graphs harmonic have herbster huang huber ieee image imprecisely information intelligence international kandola kernel kernels labeled lafferty learning lighting local localized machine making martinez mathematical mathematics matrix method methods metrics minimization model monte more morphable nato nearest neighbor neural nonparametric novel object occluded online only over pages pairwise partially path pattern perform performed person phillips poggio points pontil preliminary problems proceedings processing proposed purpose real recognition recognizing references regional regression relevance remarks research results robust roth samaras sample scholk sciences second segmentation semi series sets setting show similarities single society soulie spectral spherical statistics subset supervised synthesis synthetic systems table templates tenth test texture that then theory this transactions transductive transforms twenty umist under unknown using utilizing variant verify versus virtual vision volume wainer wechsler weston will with world yeung zhang zhou zoller
http://doi.ieeecomputersociety.org/10.1109/CVPR.2006.50	102	Affine invariance revisited Evgeni Begelfor Michael Werman School of Engineering and Computer Science The Hebrew University of Jerusalem 91904 Jerusalem, Israel	absil acad accomplish acta action advantage affine affinepermutation after airplanes algorithm algorithmic algorithms almost although amplitude anal analysis angles apllications appl applicandae applied approach arias assume asterisque average averaging barden based basic basis berger berlin berthilsson between birkuser bjork boston buser cambridge carmo carne caused changes classification clearly clusters color comp comparison computation computer computing conference constrains corresponding corresponds curvatures cvpr denote depth differential distance distances downloaded edelman editors eijv eilv enough error estimation every exist experiment failed feature figure finally flat forsyth from function geometric geometry gives golub gool grassman grassmann gromov grows hall icpr ieee illumination image images inferences inner intell international invariance invariant karcher kendall kinetic labeling left linear link mach mahony manifolds math mathematicae matrices matrix means meanshift measurements methods mindru misclassification model modern moons mordfjordeid motion moura mundy must natl natural need noise nonlinear normalization norway notice numerical ometry orthogonality outdoor pages panoramic partitioned pattern pennec performs photometric planar plot point points ponce possible prentice press probabilities probability proc proceedings processing product ratio recognition references result riemannian right rotation same sectional sepiashvili sepulchre series session sets shape show shows siam signal similarity small smith society sophus space spar special springer sspr standard statistical statistics structure subspaces successfully such surfaces symmetry task that then theory there thus tools trans ueij ueijv uekj unique using view viewpoint vision volume weinshall well were werman wiley with wong workshop worst zisserman
http://doi.ieeecomputersociety.org/10.1109/CVPR.2006.21	75	A Design Principle for Coarse-to-Fine Classification	additional adjusting advances aerial after against aggregate also alternatives although amit among andconquer andy annalsofstatistics application applications applied assign asymmetric avenues average baker baluja based basic because blanchard boosting brown buildings burl candidate cascade cation changements chicago class classes classifi classification classifier classifiers coarse coherent comparable compare complexity composition computation computational computer conclusion conducting conference constructed contains corresponding corresponds cost curve cvpr data dataset deciding decision decomposition designs detecting detection detectors different distance divide divided doing driven each ecole efficient elad energy error estimation european eveland everything examination examining example expanding experiments explanations face faces factors false feature features fewer fine fleuret focalisation france freeman freund from gavrila geman general generalization girosi given group have hierarchical hierarchies hierarchy high holistic hopkins hypotheses icml icpr ieee image improvement improvements include information intermediate international johns jones journal jung kanade keshet known krempp larger learned learning lecun letters line linear lower machine machines manbeck manual manually marchette matching mathematics maximal mcclure measure method methodology might miller model models more much mullin multi multiclass multiple murphy namely naquet nature nayar network neural neuroscience nips node nonetheless notably noted numeric object objective objects obtained once ongoing optimal osadchy osuna other ours over pages pairs pami paris parts patchwork pattern performance perona pivotal planar plays plotted poggio point polytechnique pose positive positives possible potter power powerful press priebe primary principle priors problems procedure procedures proceedings processing promising proposed quantity quaterly rank rapidobjectdetectionusingaboosted rate recent recognition reconciling reconnaissance references rehg rejection rendering report represent research results reusable role rowley sahbi sali same schapire search searches searching sequential sequentiallearningwith several shape sharing should show significant similar simple skewed slightly society socolinsky springer statistical statistics step stereo strategies strategy subject subset subsets substantially sung support synergistic systems table taken technical template test tested testing thanthesimplebinaryedgevariablesutilizedhereandtraining that their theoretic theory these thesis this tilts time torralba training transactions transform trouve ullman university using varying vector verlag versailles versions very vidal viewbased violaandm vision visual ways well with work york
http://doi.ieeecomputersociety.org/10.1109/CVPR.2006.100	57	Dimensionality Reduction by Learning an Invariant Mapping	academy adaptive advances algorithm also although analysis andheadingfromimage arbitrary architecture arts avoided avoidscollapseto belkin bengio between bottou bromley butnoneproducesafunctionthatcanacceptnew cambrige camera chapman chopra clustering component computation computer conference connections constantfunction contrastive cover cowan creatingadimensionalityreductionmappingusingprior cuts cvpr data delay desilva dimensionality discriminatively dissimilarpairs distant donoho drlim editors effect eigen eigenmap eigenmaps embedding equilibriumin evenly experiment experimental extensions face faithfullymapsnew feasible first forhighdimensional framework from function generic geometric given global grimes guyon hadsell have hessian hill huang icml ieee ifthisisnotthecase image images information input inputs intelligence interconnectedspringsdoes international invariance isomap joint jolliffe jordan july kernel kernels knowledge knowledgehas langford laplacian learn learned learning lecun lies lighting limitation limited linear locally locallylinearembedding london loss machine maintains malik manifold maps matrix meaningfullocationsonthe mechanicalsystem methods metric much muller multidimensional national negates neighborhood neighbors network networks neural niyogi nonlinear norb normalized object only other output pages paiement pami parameterized pattern pose position positions possible power press principal prior problem proc proceedings processing recognition reconstruction reduction references relationships results robot roweis sackinger sample samples sampleswithoutrecomputationorpriorknowledge saul scaling schoelkopf scholkopf science sciences seen segmentation separate sequences shah siamese signature similar similarity smola society space spectral springer strength success support system systems techniques temporal tenenbaum tesauro that theexperimentswithlleshowthatlleismostuseful thenllemaygivedegenerate this thrun time transactions twenty unseensamples uses using value vector verification verificaton verlag very vincent vision weinberger weiss wellregistered were where which with withapplications york
http://doi.ieeecomputersociety.org/10.1109/CVPR.2006.96	11	Deformation Modeling for Robust 3D Face Matching Xiaoguang Lu and Anil K. Jain Dept. of Computer Science & Engineering Michigan State University East Lansing, MI 48824 {Lvxiaogu, jain}@cse.msu.edu	acad accuracy acheroy across active addition adequate algorithm along also although analysis animation appearance applications approach authentication automatic automatically avbpa baker based being besl beumier blanz bookstein bowyer bronstein built capabilities challenge chang changes chiaki chua cloning colbry combining computational computer computing conclusions conducted conference contact contains control correctly correlation corresponding cost cover current currently curvature cvpr data decomposition deformable deformation deformations deformed demonstrate depth desirable diego discussion done driven dryden each errors examples experimental experiments expression extracted extraction face faces facial feature features fiducial fitting flynn formation formulated framework from frvt fully function future gallery generate generic geodesic geodesicbased gesture given gordon grand grenoble gross group guildford have hecker hoffman http human icassp iccv ieee ikeda image images implementation improves incorrectly integrating international invariant jain john kimmel labeling landmark landmarks large laser learn learned lischinski little local make manifolds manually many mardia marques matched matching matlab matthews mckay member meshes method milios minimization minimum minolta minoltausa model modeling models more morphable multiview natl needed neumann neutral nonrigid offline only optimized overview pages pami paths pattern pentium performance performed person phillips photographs pighin plate point popovic population pose possible practical presence principal proc proceedings process proposed pursued range realistic reason recognition reduce reducing references regions registration results robust salesin same scan scanner scans scheme scruggs seconds sethian shape shapes shows siggraph signature simultaneously small society sons southampton specific spherical splines stage statistical subject subjects sumner surface surfaces synthesize synthesized synthesizing system szeliski takes tanaka template templates test texture that these thin this training trans transfer transferred triangle type using variations vendor verification vetter vision vivid volume warps which wiley williams with without worek work would
http://doi.ieeecomputersociety.org/10.1109/CVPR.2006.106	128	Edge Suppression by Gradient Field Transformation using Cross-Projection Tensors	adaptive addition additional adelson affine agrawal align america anal analysis anandan another antipolis applications applied approach area artifacts assumption aubert avidan background balancing barrow based basri belhumeur berlin beyond black blake boult brockington calculus calibration cannot characteristics chen coherence coincide color common comp comparison components comput computer conceptually conclusions conf conference containing context create cross cvpr cybernetics davis dependent deriche derived deriving designing determining different differential diffusion domain drew easily eccv edge edges editing effects elgammal enhancement enhancing equations estimating european exactly exchange exposure extracting extraction fang farid fields filtering finlayson fixed flash focus foreground framework from funt fusion future gangnet germany glass global gradient graph graphics grimson handle handled harwood have hence hordley horn iccv ieee ikeuchi ilie illumination image images improve incorporating independent information intell intl intrinsic invariants irani jacobs jahne june kiryati kornprobst lack lambertian land layer layers lecture levin lighting lightness local machine make manipulations mathematical matsushita mccann might mixture model models multiple multivalued nayar need neighborhood nice nishino nonlinear normalization notes npar object objects operations optimistic page pages parametric partial pattern pdes peleg perez photography piccardi poisson polarization presented preservation problems proc proceedings processing projection raskar real recognition recovered recovering references reflection reflections regularization reliable remaining remains removing repeat requires result results retinex review sakauchi sampling sarel saturation scale scene scenes schechner scheme science sciences scientific seamless search separating separation sequences shading shadows showed similarly simple slight society sophia space spatio special specular springer stauffer stitching strong structure subtraction such suppressing suppression surrealism surveillance systems szeliski techniques temporal tenenbaum tensor tensors texture that theory thesis this through time times tinge tracking trans transformation transparency transparent tschumperle understanding universite used using usual valued variance variations vector verlag video vision volume weickert weiss which while white will with without work zomet
http://doi.ieeecomputersociety.org/10.1109/CVPR.2006.288	106	SpatialWeightingforBag-of-Features MarcinMarszalek CordeliaSchmid INRIA Rhone-Alpes, LEAR- GRAVIR ^ 665 av de l'Europe, 38330 Montbonnot, France Marcin.Marszalek@inrialpes.fr Cordelia.Schmid@inrialpes.fr	achieved acknowledgments adapted affine agarwal alpes also andp appearance applying applyingthe arregui auer automatic awan background based belongie best between boost boosting cambridge caputo categories categorization categorizing category challenge chapelle chung class classes classifcation classification classificationrateachievedbyourmethodonthe classifier clutter combined community computation computer conditions conference could create csurka cues curie cvpr dance darrell data decrease demonstrated depth derivative descriptors detect detection detector detectors direct discovering discriminant discriminative distanceasametricforimageretrieval distinctive dorko earth eccv efficient efros eklundh entropy european evaluation everingham excellence expected experimental extension feature features fergus first focus foreground form forobject forobjectrecognition fowlkes framework freeman fritz from further fussenegger future garding generic gool grant grauman graz guibas guiding haffner have hayman histogram hypotheses iccv icme ieee ijcv image images implicit improves incorporates influence inria integrating interest introduced invariant iwlavs jing kernel kernels keypoints lazebnik learning learningwithkernels leibe leonardis lindeberg local localization location lowe lower machines making malik maps marie marszalek match material maximum mercerkernels method methods mikolajczyk model models more mover network nine object objects obtained opelt operators optimizationandbeyond other outperforms pages pami paper part parts pascal pascalvocchallenge pattern perona pinz point ponce possibility preliminary press proceedings produced project promise proposed pyramid real recognition references region regularization relations report representation representative research results retrieval rhone robust roth rubner russell scale schiele schmid scholkopfanda scia segmentation selection sets shape show shown significance sivic slightly smola society sparse spatial spectralgroupingusingthenystrommethod state statistical study summary supervision support supported table technical technique tector texture that their this thus tomasi under unsupervised uses using vapnik vector vectormachines vision visitor visual volume weak weighting weights which willamowski williams with withlocal work workshop world zhang zisserman
http://doi.ieeecomputersociety.org/10.1109/CVPR.2006.40	83	AShape Representation for Planar Curves by Shape Signature Harmonic Embedding	abbasi academic active addisonwesley advanced affine algorithm algorithms analysis analytical anisotropic applications approach artificial based basic basri belongie biederman brandt brooks buzbee calculus cambridge chellappa classification closed common company complex computational computer conf conference conformal contexts contours correspondence cremers curvature curvaturedependent curve curves cvpr databases deformable dekker dependent descriptor descriptors determining differential diffusion diplaros direct distorted dryden dubuisson duci dynamic edition embedding equality equation equations ericsson eriksson estep european evolution expo fast faugeras flow formulations fourier framework francisco freeman from front fronts galun general generalized generic geodesic golub gonzalez gorelick graphics grimson hamilton hamiltonjacobi hansbo harmonic horn human hung ieee image indexing influence integrals intelligence intl into invariant iterative jacobi jain john johnson jolly journal june kaplan kimia kittler klein kluwer kohlberger kunttu large last lepisto letter level leventon machine mackworth malik mapping marcel mardia marsden matching methods milios mitter models mokhtarian mpeg multimedia multiscale mumford nielson nonlinear numerical object occluded optical osher pattern petrakis physics planar poisson press problems proc proceedings processing programming propagating propagation publishers publishing puzicha rahamaa recent recognition references representation research retrieval review scale schn schunck schwarz sebastian segmentation sethian shading shah shao shape shapes sharon shock siam signal simchony similarity soatto society solution solving space speed standardization statistical statistics super system systems template theory therory trans trigonometric understanding uniqueness university used using variational vemuri visa vision wavelet wiley with yang yezzi zhang zhong
http://doi.ieeecomputersociety.org/10.1109/CVPR.2006.238	145	Real-time Image-based Guidance Method for Lung-Cancer Assessment Lav Rai1, Scott A. Merritt2 and William E. Higgins1	ability accuracy acknowledgments addition airway airways alvey amini amir analysis analytic ando angioscopy anne ansar applicability application applications applying armando aspiration assessing assessment assist assists augmented automated available based because becker besser between biopsy bolles boosted bricault bronchofiberscopic bronchoscope bronchoscopic bronchoscopy cambridge camera cancer canio care cartography cascade case changes chen chest chin cinquin cleary clin clinical clough colonoscopy combined comm computer conclusion condition conf conference consensus considerable continuously convergent corner coronal ctbased ctderived currently cvpr cystic daniildis data date deformable deguchi delivery derived detailed detecting detection detector determination diagnosis dimensional directing display distance doing done drug duplaga duploga during early edge edition elecromagnetic electromagnetic emphysema enabling endoluminal endoscope endoscopic endoscopy engel enjoji ernst error estimation european extensive factors fast features ferretti fibrosis fields fischler fitting flexible frame from function further future galloway geometry globally good goodman grants graphics greatly greenlee guidance guided hager haralick hardware harmon harris hartley health helferty hence herth hewer higgins hilar hill horaud human ieee illumination image images imaging important inexpensive information institute intelligence intensitybased interbronchoscopist interest involve jones july june kenney keypoints kitasaka kussin leboullenx light linear lines live location lung lymph machine makes manduca manjunath mapping matching maurer mcadams measures mediastinal medical mehta method minami mjolsness model mori motion multidimensional multiple murray narrowed national natori navigation needle nevel nodes nolle nomura none note number object only onto operates orens orientation ottenberg outside pages paradigm partially pattern pelvis permits perspective physician physiology point points pose position possible postregistration prcoedures press problem procedure proceedings processing projection proposed provided quan random rapid rates reach real reality realize recognition references regions registered registration renderer rendering research respiration robustness roentgen rohlfing rois russakoff sakai sample scale schwarz second sensor sept shimokata shoji show sihoe similarity simple sites slice society solomon solution solutions space spie stable staging standard statistical statistics stephens suenaga sugiyama supported surface surfaces surgical system systems takabatake taylor technologies tests texture that these this three thun time tomasi tool tools toriwaki track tracking trans transbronchial transformation transverse treatment trials triggs turcza under underway university updated using variability video view views viola virtual vision visual visualization wall wang white wide wiener will with work zisserman zuliani
http://doi.ieeecomputersociety.org/10.1109/CVPR.2006.220	110	Perception Strategies in Hierarchical Vision Systems Lior Wolf Stan Bileschi Ethan Meyers Center for Biological and Computational Learning The McGovern Institute for Brain Research Massachusetts Institute of Technology Cambridge, MA	achieved acknowledgments across additional advances affiliated affine ahissar analysis appearance approach artificial attention automatic bart bartlett based bayesian berg beyond bieschi bileschi biological biologically blanz bmvc books bottou brain breiman british bullier cambridge canny categories categorization center central classes classification cognitive combined common comp company component computational computer conceptual conf conference conte contextual contract corporate corporation correspondences cortex cotext crcns criepi cross csail cvgip cvpr daimlerchrysler dalal darpa demonstrate dept describes descriptors detection dickinson different discriminative distortion done dscovery easier eastman edge elaborate electric employing eugene even evidence example examples experimental face faces fahle feature features feedback fergus first fize forests form foundation fourteenth fragmentbased framework from fund gain generalization generative generic gestalt gesture global gradients grants harder hawkins health heisele helpful here hierarchies histograms hochstein honda huang human iccv identification ieee ijcv illuminations image imaging implication implicit incremental industries industry inspired institute institutes integrated intelligence intl invariance involving journal july kanade kind kodak komatsu laboratory lazebnik learning lectures lecun leibe leonardis less lighting limited local london loops lynch machine machines malik marlot marr matching mcdermott mcgovern memorial memory merrill metal methods model modeling models modle morphable motivated motor much nakayama national nature naval networks neural neuron neurophysiology nips nishihara nonlinear novel object objects office oram organization orientations oriented oxygen pages pami pattern pentland perception perceptual performance perona perrett pictures poggio ponce pose poses possibly potter power present press priming probabilistic proc proceedings processing provided psychology quartly random recog recognition references regions relationships replacement report representation representations research reverse reviews robust romdhani rosenfeld royal sali schiele schmid schneiderman science sciences search segmentation series serre shape shapes short siemens single society sony sparse spatial speed sponsored stacked stadard stat strategies studied submitted sumitomo support system systems term tested texture than that thing this thorpe thought thrun times tong torralba towards toyota training treisman triggs ullman underst unified using vector verification versus vetter view views vision visual volume volumes weber well welling which with wolf wolpert workshop
http://doi.ieeecomputersociety.org/10.1109/CVPR.2006.232	108	Derek Hoiem	about above academic acknowledgements adaboost advances after along always ambitions annotation aspects autocalibration autonomous award barrow based baseline bayes bayesian become being biederman bill binford blue boxes bregman brooks building camera capable career carreira cases causes challenges chapter characteristics characterization chen classification classifiers code collection collins color comaniciu comparisons compass complete computer computing conditional conf conference confidence considerably consistency context contextual correctly could course cvpr cyan dalal data database depth detected detection detections detector dimensional discriminative distances dotted dual early eccv editors efros elsevier erlbaum estimate estimated estimates estimation euclidean eventually example face false field fields figure forsyth found framework freeman from fulfill full funded geometric geometry glance global gradients graphical green greienhagen greiner ground grow guaranteed hallucinated hanson hashimoto hebert hierarchical high histograms hoiem hope horizon http human iccv identified ieee ijcv illustration illustrative image images improvement improves inference influences initially intell intelligent interaction interpretation interpreting into intrinsic isie jeong joint jones kanade kaufmann knowledge kosecka krahnstoever kubovy kumar labeling labelme lane large lawrence learning likelihood line lines local logistic lower machine machines mahler make margin matlab mendona methods middle missed model modeling models more morgan moshe most multiscale mundy murphy natural navneet network networks niemann nips object objects often ohta oliva organization oriented outdoor outputs pages pami parsing part particular parts pascal pattern pearl pedestrian pedestrians perceptual performance perpin pitman places platt plausible pomerantz press priming probabilistic proc proceedings providing publishers ramesh random real reasoning recognise recognition recognizing recovered recovering references regression regularized removed repeated report research researchers restricted results riseman road robust rothwell russell saliency scale scene scenes schapire schneiderman science segmentation semantics sense show shows singer single sinha society solid springer standpoint statistical statistics structure sudderth suggestions support surveillance system systems takeo technical tenenbaum thank that they this three time tool toolbox torralba triggs true types uncalibrated understanding unified unifying useful using usually vector vehicle verlag video viola vision visions volume website where will wilsky windows with worse would wrong yellow yuille zemel zhang zisserman
http://doi.ieeecomputersociety.org/10.1109/CVPR.2006.216	19	Xiaoming Liu1	actual advantage alexander allinson along analysis appearance application applications applied assisted attention automatic baddeley baker based basis belhumeur best better biometrics blackburn blanz bone bruce bsat california camera canonical chase chaudhuri chellappa chen class classification cognition cognitive computer computing conf conference construction contributions cvpr czech database davis depend dependency detection diego dimensional direction distribution domain duda each eccv ecological edition effect effects eigenfaces einhuser erlbaum especially evaluation exposure expression face faces feature fields figure fisherfaces fitting found france from frvt gaze geometry gesture graham grenoble gross grother hand hart have heads help hespanha hillsdale hope however human identification ieee illumination instance intelligence interaction interesting into intrinsic invariant japan john journal knowledge korea kriegman lane laugherty learning light lighting likelihood linear literature logie long machine malsburg many matthews mechanism methods micheals might model modeling morphable morphing multi multiplying nara neuroscience number object objects obtained okada once opportunistic optimal other pages palmer paper parametric pattern pentland perception performance perona personal perspective phillips photograph pose position practical probabilistic problems proc proceedings projection psychology reassessing recognition references reflectance relevant report republic research robust rosch rosenfeld seoul shape single smiling society sohn sons space specific stability stimulate stork subspace subspaces such surface survey tabassi target test themselves there these this three time tion toole topics training trans transformation true turk type understand unfamiliar using valentine validity values vendor vetter view viewpoint views vision volume washington weber weinshall welling werman which wiley will with woodhead workshop yacoob york zhao
http://doi.ieeecomputersociety.org/10.1109/CVPR.2006.163	119	Landmark-based Geodesic Computation for Heuristically Driven Path Planning	account agrawala algorithm beers chaddha compressed compression computed computer conclusion conference cvpr data dimensional distances extraction fast from geodesics have heuristics higher ieee images into landmarks marching memory modification pages paper pattern presented proc proceedings recognition reduce references rendering requirement scheme siggraph simple society specific speed structures surfaces takes textures this triangulated using vision
http://doi.ieeecomputersociety.org/10.1109/CVPR.2006.42	20	A Visual Vocabulary for Flower Classification Maria-Elena Nilsback and Andrew Zisserman Robotics Research Group, Department of Engineering Science University of Oxford, United Kingdom	abbasi achieving acknowledgements affine already also analysis appearance application articulated aspects associated aster bags based bayesian been bluebells boost bosch botany bottom boundary boykov bray categories categorization catergories challenges changes class classi classification classified classifier classifying closest code collections column combining computer conference contain coping correctly could csurka curie cuts cvpr dalal dance database depth descriptors detection detector detectors developed difficulty discovering distance distinctive dorko each eccv efros elpel employing epitomic example examples explicit features fication figure final flower freeman frey friends from funded gatica gave general gool gradients graph have having hebert hierarchical histogram hops human iccv identification ieee ijcv illumination image images indeed independence individual inner inria instances instead interactive interest intersections invariant irises jacobs jojic jolly kannan kernels keypoints kumar latent lazebnik leaf learning least left level ling local lowe mahamud many marie marszalek match matching materials measure middle mikolajczyk model modelling models mokhtarian monay more munoz natural network number object objects odobez ones optimal oriented pages parts patiently pattern pawan perez performance perona petals plsa point press principal proc proceedings processing project providing purpose quelhas recognition references region report results rhone right russell scale scaleinvariant scanned scene scenes schmid segment segmentation selection self several shape shapes shown shows significant single sivic society specialized springer stamen statistical study such sufficient superior technical test texture thanks that this tigerlilies training transactions triggs tuning tuytelaars using varma varying verlag viewpoint vision visiontrain visual vocabularies vocabulary volume where with work workshop zhang zisserman
http://doi.ieeecomputersociety.org/10.1109/CVPR.2006.77	91	Color Subspaces as Photometric Invariants Todd Zickler Satya P. Mallick David J. Kriegman Peter N. Belhumeur	acknowledge acknowledgment albedo algorithms america analysis anandan angelopoulou applications approach approximations authors barnard based belhumeur beyond black bolle boykov british cardei central chellappa chromaticity class code color comparison components computational computer conditional conf conference constancy cremers curless cvpr data dimensional direction drew efficient estimation european experiments fields finlayson florida flow foundation framework from funded funt generously gevers group hancock hertzmann highlights hordley ieee ikeuchi ikeutchi illuminant illumination image images implementation intelligence intensity internet invariants inverse iterated jongwoo journal kanade klinker kolmogorov kriegman lambert lambertian light like machine mallick markov material measurement methodology michael modes motion multi narasimhan national nayar nishino nonwhite object optical pages park part passive pattern photometric proc proceedings processing provided ragheb ramesh random ratios recognition reconstructing references reflectance reflection released removal removing representation request research robust science sect segmentation seitz separate separating shading shadows shafer shape shedding single soatto society space specular specularity spie stereo stereoscopic structure supported surface surfaces synthesized textured their this three through trans under unifying university upon used using varying veksler videos view vision vladimir volume weijer were with wolff work would yang yezzi zabih zhang zheng zickler
http://doi.ieeecomputersociety.org/10.1109/CVPR.2006.14	14	3D Facial Expression Recognition Based on Primitive Surface Feature Distribution Jun Wang, Lijun Yin, Xiaozhou Wei and Yi Sun Department of Computer Science State University of New York at Binghamton, NY, 13902, USA	above acknowledgement acquisition action actions active addition aided akamatsu algorithm analysis application approach approximating automatic bartlett based behavior binghamton bouguet bowyer braathen budynek bulletin canada capture capturing challenge chang chellappa chen chiaki classification classifying coding cohen cohn computer computing conf conference consulting context correlation cosine cosmos cowie cubic current curvature curves cviu cvpr cybernetics data database davis department design diego digital direction directions discrete document donato dorai douglas dynamic editors ekman elgammal emotion essa eurographics expression expressions face facial farin fasel fellenz final flow flynn form foundation francisco frank free friesen from fusion future garg generalized geometric girod give gokturk goldfeather gong grand grants graphics gray hager haralick high hoffman http huang human iccv ieee ikeda image images improve independent information integrating intensity inter interaction international interpretation interrante investigate investigator issues jain james kanade kaufmann kollias laffey lainscsek learning literature littlewort long luttin lyons machine magazine manifolds maps marques material measurement mechanisms metaxas modelbased modeling montreal morgan motions movellan movement national neurological neuropsychology novel nystar objects optical order overview pages pami pantic part pattern pentland phillips planning press principal probabilistic proc proceedings processing producing program psychological psychologists purpose recognition recognizing references report representation research resolution review rinn rise rosato rosenfeld rothkrantz samaras scale scheme science scruggs sebe sejnowski sequences shape signal single smith society southampton space spherical splines spontaneous state static subtle suny supported surface surfaces survey surveys synthesis system tanaka taxt taylor technical technique temporal texture third this tian tomasi topographic tracking trans transfer transform trier tsapatsoulis turk under understanding units upon using vectors video view views virtual vision votsis wang washington watson will worek work workshop yacoob zalewski zhang zhao
http://doi.ieeecomputersociety.org/10.1109/CVPR.2006.177	9	Local Steerable Phase (LSP) Feature for Face Representation and Recognition ZHANG Xiaoxun and JIAYunde Department of computer science and engineering Beijing Institute of Technology, Beijing 100081, P.R. CHINA {zhangxiaoxun, jiayunde}@bit.edu.cn	abdenour acknowledgements adelson ahmad ahonen algorithms allan also alternative analysis appealing appearance applied approach april audio authentication based boosted boosting canny carneiro cascade chengjun china classification classifier classifiers computationally computer conclusions conference correlation correspondence criteria cvpr cybernetics daniel darabiha database databases david deal dense depth descriptor design detection discriminant discriminate disparity edward efficient encode encouraging enhanced estimation europen evaluation experimental extended face feature features feret filters fisher fleet foundation fpiv frame frankfurt freeman from further gabor germany goethe grant gustavo hadid hardware harry heyonjoon hyeonjoon ieee image implemented information institute intelligence internatinal internation international issues jacob james jepson johann jonathan jonathon jones journal kittler krajsek like linear local luettin machine maclean maenpaa main maitre march matas mathews matti measurement merl messer methodology michael model moon more motion multiple national natural nearest neighbor object other partially patrick pattern paul peitikainen performance person phase phillips physics presented problem proc proceedings processing programmable protocol rapid rate rauss recognition references regions report reports representation response results richard rizvi rose scale scharstein science show simple society space stan statistical steerable stereo supported syed systems szeliski table taxonomy technical this timo topi trans typical universitat unser used using verification video viola vision vtsdb washington wechsler weighed weighted william with wolfgang work workshop zhang zhiyi
http://doi.ieeecomputersociety.org/10.1109/CVPR.2006.119	26	Fast Human Detection Using a Cascade of Histograms of Oriented Gradients Qiang Zhu1, Shai Avidan2, Mei-Chen Yeh1, and Kwang-Ting Cheng1	accuracy achieved algorithm analysis appearance applied approach approaches aspect assembly based belongie beyond blocks boosted cambridge cartesian cascade cascaded case central centric chosen coference comparable comparing comparison compute computer concept conference contexts cviu cvpr dalal dense density detecting detection detector detectors different distinctive down eccv efficient either error especially european evaluate evaluation experiment extract false fast faster fastest feature features felzenszwalb figure figures filter finding forsyth fppw from gavrila goes good gradients higtograms histogram histograms human huttenlocher hypothesis iccv ieee ijcv image images implementation increases integral integrating intelligence internation international invariant ioffe jones journal kernels keypoints large larger learning level locations lowe machine machines matches matching method methods mikolajczyk minimum miss more most motion movement much multiple norm norms note object optimization oriented over pami papageorgiou part pattern patterons pedestrians people performance philomin pictorial pixel poggio porikli positive post press previous probabilistic proceedings processing produce puzicha random rapid rate rates ratios real recognition recongition rect rectangular references refers regularization rejectors reported required results robust sample scale scan schmid schneiderman scholkopf selected shape simple size sizes smaller smart smola snow society spaces sparse speedup step structures success support survey system table testing than that their this time total trainable transactions triggs understanding used using vector vehicles viola vision visual visualizing when window windows with zisserman
http://doi.ieeecomputersociety.org/10.1109/CVPR.2006.274	126	Shape from Dynamic Texture for Planes	acknowledgments advanced affine after agency algorithms analysis anandan application artificial autocalibration autocorrelation award axis based bayesian behavior bergen bose boston both bottom british brown calibration cambridge camera case characteristics chiuso chou computer concurrently conference coordinate cremers criminisi crowds curved cvgip cvpr data dense developed direct direction directions directly distortion doretto dynamic dynamics eigenvectors error escalators estimates estimation european expected favaro figure finally first fitzgibbon flow foreshortening form frames from future general geometry gibson gradients graphics grimson ground hancock hanna hartley have hierarchical highway highways hingorani homeland homogeneity homogeneous horn houghton human iapr ieee ijcai image including intelligence interesting international investigate isotropic isotropy iterative jones joseph journal kanade kanatani krahnstoever learning line lischinski local location lucas machine magnitude malik many marked mendonc methods metric mifflin mixing model modeling motion movie moving multiple nelson nevatia noise norm normal norms nowhere objects optical orientation paper partially patten pattern perception periodic pets picard pixel plane point polana port press principle proceedings processing project projective promising pursued qualitative ravichandaran recognition recover recovering rectification references registration relates relating research revisited ribiero rigidity rosenholtz scenes second security seems segmentation self sequence shape shvaytser soatto society spatial spatially static statistical stereo stochastic strength supported surface surfaces surveillance synthesis szummer technique temporal test texture textures that there this tolerance tracking transactions truth understanding university upgrade using utilized values vanishing variety vertex vidal video videos view vision visual walking weldon werman with witkin workshop world yaniv zhao zisserman
http://doi.ieeecomputersociety.org/10.1109/CVPR.2006.203	87	Multiscale Nonlinear Diffusion and Shock Filter for Ultrasound Image Enhancement	ability acton adelson alvarez anal anisotropic arridge aubert biomed blurry boundary brockett burt capability catt code coherence coll commun compact computer conclusion conference continuous coulon coupled coupling cvpr demonstrated deriche detectabilities detection diagnostic diffusion dual echo edge edges effectively elmoniem engr enhance enhancement equations evolution feature figure filter filters from icassp icip ieee image images imaging improve intell kadah kornprobst laplacian lions machine malik maragos mazorra measurements method miua morel morphology multi multiscale noise nonlinear novel numer oriented osher pages pattern performance perona phantom potential probabilistic proc proceedings processing proposed pyramid quantitative real recognition reducing reduction references removing restoration results rudin scale selective shock siam signal small smoothing society space speckle spectral structures studies time trans ultrasound using vision vivo while with youssef
http://doi.ieeecomputersociety.org/10.1109/CVPR.2006.169	58	Learning Joint Top-Downand Bottom-upProcessesfor 3D Visual Inference	able about accv acknowledgments activity agarwal algorithm allan also alternative andb annealed anonymous appearance applied approach approximations arms around articulatedbody aspects augment automatic available avoiding background banff basis bayesian been bengio beyond bhatia bishop black blake body bottou capture captures cast centered chair cham clutter clutterandsometimes cluttered combining comes comments common complex computation computer conditional conference configurations continuous contrast cristian crowded currently cvpr dalal darrell database dayan deformable demirdjian density detecting detection deutscher different difficulties dimensionaland directional discriminative discussions distinctive document dynamic dynamical eccv effect efros either elgammal embedded encodings enhancement environments especially estimated estimating estimation experts exploring factor fast feature features figure filmed filtering fitting fleet foreground framework frey from generative ghahramani gradientbased gradients grant graphical grauman ground haffner hashing hertzmann hierarchical high hinton histograms homogeneous html http human huttenlocher iccv icml icondensation ieee ijcv illumination image images important includes ineffective inference inferring informative insightful invariant investigate isard jacobs jepson jojic jones jordan junction kanaujia kernel keypoints kumar layered layers learned learning learningflexiblespritesinvideo lecun leibe level likelihood limb limbed linearly local localization long loose lowe make malik manifold maps metaxas methods metric mixtures mocap modeling models modes monocular more mori motion multilayer multiple murphy natural neal networks neural nips noisy notice noticethedifficultyoftheposesthatinvolveself objects occlusion office online oriented other overall overlap pages parameter partial particle partly pattern patterns pavlovic pedestrian pedestrians people perfect plan pose poses position posture precisely press priors probabilistic proc proceedings promising propagation provided pursuit qualitative ramanan recognition reconstruct reconstruction recovered recovering recovery references regions rehg reid representation reviewers robust rosales roth scale scene scenes schiele science sclaroff search seeman segmentation selection sensitive sequence sequences sets severalreconstructionfromthedatabaseof shadows shakhnarovich shows sidenbladh sigal silhouettes sleep small sminchisescu snow society some somewhat sources space spaces specialized state statistics still stochastic streetlight strong study subjects subtraction supported svensen switching taycher term than thanks that there this torr toyama tracking training trees triggs truth under unifying unsupervised urtasun using valuable variational video viola vision visual volume wake when windows with without zisserman
http://doi.ieeecomputersociety.org/10.1109/CVPR.2006.38	82	A projective invariant for textures	above achieving affine after analysis applying approach argument automatic avnir azencott based blanc both changing characterization classification classifying complete computer conci conclude conference constant corresponding could cover covered cvpr definition denote denotes diameter dimension each enough equals espinal extended falconer field filters first following form fourier fractal francisco freeman from gabor geometric geometry global good gool hand hartley have heitz hosticka icip ieee ijcv illumination image images independence inequalities invariance invariant invariants jawerth john kaplan kervrann kubota laine lazebnik less letters level limit lipschitz local logc logn logr mandelbrot markov matched materials measures mihran mindru modelbased moment monteiro moons most multifractal multinomial multipermuted multiple mundy naor nature need number object obtain order original other packet pami pattern peleg pichler ping point ponce press proceedings process proof prove random recognition references regions representation resolution same schmid segmentation sets show sides signature signatures similarily smallest society sparse spatial statistics such suppose taking talon target techniques teuner texture than that then theorem these thus trans transform transforms tuned tuytelaars under understanding unsupervised using varma viewpoint vision wavelet weiss where which wiley windowed with written yields younes zisserman
http://doi.ieeecomputersociety.org/10.1109/CVPR.2006.48	48	Activity Analysis in Microtubule Videos by Mixture of Hidden Markov Models Alphan Altinok1 Motaz El-Saban2 Austin J. Peck3 Leslie Wilson3 Stuart C. Feinstein3 B. S. Manjunath2 Kenneth Rose2 Computer Science1, Electrical and Computer Engineering2, Molecular, Cellular, and Developmental Biology3 University of California Santa Barbara, Santa Barbara, CA, 93106	accepted accuracy acknowledgements acoustic acta actin activities advances aggarwal algorithm algorithms along analysis annealed annual assessing assisted association audio automated automatic axes bahl based behavioral between beyond bicego biochim bioimage biol biological biology biomedical biomedicine biophys body bremond british bruckstein caliber cambridge capabilities cell cells cellular center chang changes chem clustering collection comparative computer computing conf conference confocal consideration contrast contributions cooperating correspondence correspondences crossings culture current cvpr danuser data datasets death design despite deterministically differential difficulty diffraction direct direction duncan dynamic dynamics edge effects electron emerged ends engineering established estimate estimation euclidean extract extraction facilitate facilitates fast feinstein findings flow fluorescence fluorescent fraction from function fundamentals funded furthering future gmbh grading grant greedy ground grouping hadjidemetriou hall have hidden higher hope human icpr ieee image images imaging improvements inability increased induces influence informatics information insights instability integrators intelligence interference internal international intervention isolated jiang journal juang kamath kamgar kimmel kinetochore kinetochores kofahi laplacian larger leading length level liang line live loss machine macro maddox mammographic manual manually maps marching markov marti matching maximum mcewen mechanism mechanisms median mediate medical medicine method methods miccai microscope microscopy microtubule microtubules might model models most motion mtds multiframe mutual nano nanometre neuronal neurons nevatia noisy noniterative normally notably novel number objective observation offers omar optimal other over paclitaxel page pages parameters parsi pattern patterns performance platform plus point polymer prentice press previously proc proceedings processes processing properly proposed publication quantification rabiner radical recent recognition recognizers reconstruction recovery reduced references reflection regularized regulate regulatory representation required research resistance results retinal richard robust rose rubin salmon samples segmentation semantic sensitivity sequence sequences serve sethian sets shafique shah show signal similar similarities similarity society somboon spatial speckle speculation speech spindles springer standardization state structures study such support symp technology temporal that their this time tirfm tomography toomre total tracing tracking tracks tran trans transactions truth tubulin unattainable under understanding union univ upon using vallotton verlag vessel vision visualization volume wacv which will wilson with work xenopus years zero zwiggelaar
http://doi.ieeecomputersociety.org/10.1109/CVPR.2006.180	96	Measure Locally, Reason Globally: Occlusion-sensitive Articulated Pose Estimation Leonid Sigal Michael J. Black Department of Computer Science, Brown University, Providence, RI 02912	about account accounting achieve acknowledgments addition adjacent after algorithm algorithms also analysis analytically andrew approach appropriate approximate articulated assumptions automatic award bayesian belief berg between beyond bhatia binary black body bottom bregler camera capture cardboard cohen column combining common compare competing complex computer computing conclusions conf conference configurations constraints corporation covariance crowded cvpr cyclic data deutscher distributed distribution driven each eccv efros error estimate estimating estimation evidence experimental explanations explicitly exponential extension face factor favorably felzenszwalb figure figures films finding fleet forsyth frame frames freeman from functional gesture global graph graphical hastie hidden howe human huttenlocher iccv ieee igert ihler ijcv ijrr illustrate image images inference intel interactions into introduce introduced introduces introducing involves isard iterations kinematic kjellstrom learning learnt least left leonid leventon like likelihood likelihoods limbed literature local lola loops loose lower malik mandel maps marginal marginalizes match matlab mckenna mcmc mean message minutes model models mori motion movie movies multiple nevatia nips nonparametric novel object occlusion occlusions ofthe ormoneit over overhead pairwise pampas pampastree parametrized parse part partial partially parts passing pattern people performs pictorial pictures pixel pose poses priors probabilistic problem proceedings propagation proposal proposed providing published quantitatively ramanan real realvalued reasoning recognition reconstruction recovering recovery references region reid relationships representation representing resulting results ricketts right roberts ronfard roth roughly same samples sampling scaled scenarios schmid search segmentation sensitive sets show shows sidenbladh sigal similarities single situations sminchisescu society solves state static stochastic strike strong structured structures stylized such sudderth summary supported takes tend than thank that this tracking tractable tree trees triggs twists upper uses using valued variables video vision weighed well where which willsky with without work would yacoob yang zhao zisserman
http://doi.ieeecomputersociety.org/10.1109/CVPR.2006.64	29	Automatic Cast Listing in Feature-Length Films with Anisotropic Manifold Space	achieve adini affine algorithm algorithms analysis anisotropically appearance applications assumption automatic based been belhumeur believe between bichsel biometrika bishop boundaries boundary breaks caip cast changes chellappa civr class classifying clustering cognitive comparing comparison compensating component computer computing conclusions conditions cones conference consequently corresponding costume cvpr demonstrated detecting detection digital dimensionality direction each eccv eigenfaces eigenspace employ enough everingham evidence extracted extracting face faces feature film films filter fitzgibbon framework fromherz fukui full further future generate geometric georghiades global good groundhog growing hall hampapur hotelling however ieee ijcv illumination image important improvement increase inherent international invariant isotropic jaffre jain joly jones journal kankanhalli knowledge kriegman labelling langford learning lienhart lighting like listing listings literature maintain makes manifold manifolds marshall martin merging method methods miller model models moghaddam more moses motion movies multi multimedia neuroscience noise nonlinear note obtained online otsuji pami partitioning pattern patterns pentland person phillips possible present press principal probabilistic problem proceedings production projection proposed qualitatively question quotient real recognition reduction reference references relations report research results retrieval robot robotics robust robustness rosenfeld royal scene science segmentation self separate sets shot shown silva similar sivic smoliar society sophisticated space speaking spie splitting spotting statistical stucki studies subspace subspaces summary survey surveys symp systems technical tenenbaum that their them this those time tipping tonomura tools tracking tradeoff turk ullman under using variable variates varying video vieo viewpoint viola vision visual wang were weymouth which wiley will with would yamaguchi yang zabih zhang zhao zisserman
http://doi.ieeecomputersociety.org/10.1109/CVPR.2006.23	143	A dynamic Bayesian network model for autonomous 3d reconstruction from a single indoor image Erick Delage Honglak Lee Andrew Y. Ng Computer Science Department Stanford University Stanford, CA 94305	aaai abbeel about accurate acknowledgments algorithms also ambiguous analysis appearance application approach approached approaches architectural architecture artificial ashutosh atlanta attempts attribute avoidance based bayes bayesian because believe better blurred bottom boundaries boundary bradski brightness british building calibrated calibration camera campus choose chroma chung chuong cipolla classifiers color comaniciu combines comparison complex computation computer conclusion conf conference construction contain context contract coughlan criminisi cryer cues cuts cvpr darpa debevec decision dellaert demonstrated depth detect detecting detection different difficult discriminative discussions distances divergence diverse does domain down drummond dynamic edge edges efros environments errors estimation european even expectation exploit factorization failed fast favaro feature figure floor found fowlkes framework from gary generative geometric geometry geometryand give grammar graph graphics graphs grouping guided have hebert helpful heuristic high highly hoiem holds hybrid ieee image images important indicate indoor inference inferred information inherently intelligence interactive internet jackel jensen jordan kanade knowledge kosaka lagr larry learned learning level like lines logistic machine made majority malik manhattan many martin maximization mean meer method metrology michels mobile model modeling models monocular more moreover mosaics motion naive natural navigate navigation network networks neural normalized nourbakhsh number numerous obstacle obtained often ones only orientation orthography other outdoor outlier pages panoramic paper parsing parts pattern perform performance photographs pieter points possible prediction present prior problem proc proceedings processing program promise radiance reasoning reasons recognition reconstruction reconstructions recover references regression reid reinforcement relevant remains rendering research rich robertson robot robust robustly saxena scene scenes schindler sebastian segment segmentation sense shading shah shape shapes shift showed shum siggraph simple simultaneous single soatto society space speed springer stanford streams successful successfully such supported survey systems szeliski task taylor tenth test texture textured textures thanks that them there these this thrun thus tomasi toward trans transforms tsai ulrich uncertainties under understand understanding uniform using vanishing verlag view vision warm when which whose with work workshop world york yuille zhang zisserman
http://doi.ieeecomputersociety.org/10.1109/CVPR.2006.26	17	A Generalized EM Approach for 3D Model Based Face Recognition under Occlusions Michaelš De Smet, Rik Fransens, Luc Van Gool K.U.Leuven ESAT-PSI Kasteelpark 1, 3001 Leuven, Belgium	account accurate acknowledge acknowledgements active algorithm amfg analysis appearance april authors automatic baker barcelona based baseline belhumeur benavente beymer blanz cambridge capture center class compensation component computer cones conference constructing cvpr damaged data database december dempster dimitrijevic elmar error estimation example expression eyeglassless face faces facial fields fitting fpiv fransens from georghiades glasses gool gross hwang iccv icip ieee ilic illumination image images imprecisely incomplete inez institute jiang june kenmochi kotani kriegman laird light lighting localized machiraju mart massachusetts matthews maximumlikelihood method model modelbased modeling models moghaddam morphable multi multiple novel occluded occlusion pages pami parametric park partial partially pascal pattern pfister poggio pose principal probabilistic proc proceedings project rama recognition recognizing reconstruction recursive references removal report reveal rubin saito sample sequences shape siggraph silhouettes single society spain stat stereo strecha support synthesis tarres technical technology this torres uncalibrated under using variable variant variations varying vetter video view views vision wide with zhang
http://doi.ieeecomputersociety.org/10.1109/CVPR.2006.110	121	Element-Free Elastic Models for Volume Fitting and Capture Jaeil Choi Andrzej Szymczak Greg Turk Irfan Essa College of Computing Georgia Institute of Technology Atlanta, GA 30332-0280	about across active algorithm alignment allows along also among analysis animated another applications approach archives articulated artifi attributes axel baker balloons based bathe beissel belllabs belytschko berlin body braunschweig bray brostow camel capture carving caselles cheung choi cial cipolla clarkson classification cohen collision color combination combined combining compensation complex computation computational computer computermethodsinapplied computing concept conclusions conference constraints contour contours convex correspondences cost creatures cuts cvpr dataset deformable deformablemodels deformations demonstrate describe desktop detection direction directions distance does dolbow dshapeandnonrigidmotion each easy eccv effcient efficiently elastic elasticity element elementfree energy engineering enough essa estimation experimental experiments feature figure finds finite first fitted fitting four fourteen frame free freegalerkinmethod fries from fromsilhouette full functions future galerkin generate generated geodesic germany gool graph graphics gross hall hashing have heidelberger horowitz html http hull hulls human iccv ieee ijcv image images including information initial input integration intelligence interesting international intersection introduction iterations iterative jenkins journal kanade kass kehl kimmel kinematic kinematics koenderink kutulakos kwatra lancaster large laurentini least line linear local markerless mataric matching material mathematics matthies mcinerney measure mechanics meshes meshfree meshless metaxas method methods minimizes model modeling models modify more motion motions moving mueller muller multi multiple netlib nodal noise nonrigid novel numerical object objects occluding occlusions only optimal optimized overview pages pami parameter park parts pattern penalty perception performed petland points pomeranets pose potential power prentice presence present presented procedure procedures proceedings program programming properties puppet real recognition reconstruction recovering recovery references refinement report represent representation research rest results review rigid river rotation rotations saddle salkauskas sample sampling sapiro segmented seitz self sequence sequences shape siggraph silhouette silhouettes simulationofelasticityforvolumefittinginducesacomputational skeletal snakes society solid some space spamm spatial squares state steedly stereo stills stochastic structure such surface surfaces szymczak technical tell terzopoulos teschner texture that theory these time torr toward tracking transactions understanding upper used using variety videos view views vision visual visualization vogiatzis volume volumetric voronoi were what with witkin wong
http://doi.ieeecomputersociety.org/10.1109/CVPR.2006.178	12	Locally Linear Models on Face Appearance Manifolds with Application to Dual-Subspace Based Classification Wei Fan & Dit-Yan Yeung Department of Computer Science, Hong Kong University of Science and Technology	abdi achieve adaptive analysis angles appearance apply approach arandjelovic assuming automatic average based bayesian between biometrika bjorck boosted both cannot cascade chellappa chen cipolla classification compared complex component comput computation computer computing concluding conf conference consider contains currently cvpr darrell data december density detected detection dimension dimensionality discriminating discriminative divergence dual dualsubspace duda during dynamics eccv embedding empirical enhance even examples experimental explicitly exploit extracting extrapersonal face faces features final fisher fragmented framework from fukui further future gaussian generative geometric gesture gives global golub guarantee hadid hart have hence hidden higher hopefully hotelling however human ieee image images information inherent internation international intrapersonal jebara jones journal july kambhatla kriegman langford leen linear literature local locally long maeda majority make manifold manifolds markov mathematics meaningful method methods model modeling models moghaddam moreover moving multiple negative neural nonlinear novel numerical object observations obtain only outperforms pages pattern pentland performance phillips pietikainen positive presented previous principal probabilistic proc proceedings promising psychological rapid rate rates rather recogition recognition recognizing reduction references relations relationships remarks represented representing results roark rosenfeld roweis saul science sequence sequential sets shakhnarovich shot show silva simple simplistic simply sixth society some specifically still stork subspaces successfully surv survey synthesis table temporal tenenbaum term test that there they though toole training typically using variates variations very video videobased viola vision volume voting well when which while wiley will with work yamaguchi yang zhao zhou
http://doi.ieeecomputersociety.org/10.1109/CVPR.2006.57	22	Tamara L. Berg University of California, Berkeley Computer Science Division	accuracy accurate alligator allocation allowed always animal animals annotation appear appearances appears approach aspects automatic bags barnard based beaver beavers because believe berg better beyond bias blei blue blur both bottom buhmann build butforsmallersetsitcanbehelpful case cases categories category censored chimps chosen class classification classifier classifiers classify clean cleaner clustering collected collections color computer conference constructing contain containing contains content correspondence corresponding create cross curves cviu cvpr data dataset datasets depictions detecting didn difficult dirichlet discovering discriminative dissimilarity distortion documents duygulu duyguly each easier eccv edwards efros empirical enormous entire evaluation exclude excluding exemplar exemplars extended extremely failed features feng fergus figure filter find forsyth freeman freitas from gathering generative geometric gibbons giridharan good google green having hierarchical high iccv ieee image images including incredibly indexing infigure initial invariant ircing jeon jmlr joint jordan june khudanpur kinematics klakow krause labeled large larger latent lavrenko lazebnik learning lemurs likelihood lines linguistic localizing location magenta maire malik manmatha many matching measures media modeling models monkey monkeys more most much multimedia multiple nearest neighbor nips nock number object objects only other ourmonkey pages pami parts pattern penguin perform performance perona petkova phenomenon picture pictures ponce poorly poses possible powerful precision previously primates probabilistic procedure proceedings produce produced provided puzicha pyramid pytlik queries ramanan random recall recognition recovering refer references related relevance result resulting results retrieval returned rich rubner russell same samples scale scenes schmid search seen selected semantic sets shape should show shows sigir since sivic society some spatial species startlingly statistical sudderth suggest supervised template tend test tested testing text texture textured than that their them thesis this tomasi topic torralba trained training uncensored unsupervised using usually varied variety virga vision visual wang well which while whole wide willsky with word words workshop worse yanai yellow zhang zisserman
http://doi.ieeecomputersociety.org/10.1109/CVPR.2006.72	32	Body Localization in Still Images Using Hierarchical Models and Hybrid Search Jiayong Zhang1, Jiebo Luo2, Robert Collins3, and Yanxi Liu1	arms assembly automatic based bayesian berg between bhatia black blake body building cally carnegie category clothed cohen collins combining comp computer conference configurations constraints cvpr database detection detectors driven eccv efros especially estimating examples failure felzenszwalb figure finding fitting forsyth fully gross however human huttenlocher iccv icpr ideal ieee image images institute interesting ioffe isard learning left limbed localization loose maccormick malik maps mcmc mellon methods mikolajczyk mixture mobo mode model models modes mori motion nonlinear object obvious open pages pairwise parse part parts pattern people pictorial pictures pose poses preferred probabilistic problem proc proceedings proposal ramanan ranked recognition recovering references remains report result right robotics robust ronfard roth scenario scheme schmid segmentation select selection shape sigal single society sprague static still strike structures stylized sullivan technical there this tracking triggs university using vision volume where would zhang zisserman
http://doi.ieeecomputersociety.org/10.1109/CVPR.2006.275	70	Shapefrom Shading: Recognizing the Mountains through a Global View Qihui Zhu Jianbo Shi Dept. of Computer and Information Science University of Pennsylvania, Philadelphia, PA 19104	albedo algorithm anal andviscosity applicable approach approximation arrifi artifi bichsel bobick boundaries brooks bruckstein bruss carnbridge chellappa ciul comput computation computer conference coordinate coupling crouzil cryer cvpk cvpr descombes deterministic direct direction dupuis durou editors eikonal equation estimating estimation fast faugeras from global hawaii height horn iccv ieee ijcv illuminant image imaging improved imuge inrelligence intelligence intensity interactive iunding journul july june kimmel leclerc light linear lkeuchi lrnage marching mathematicul matlab matsushita merno methods methoils mjtai multiresolution numerical nzrrnel occluding oliensis oprimiution optimal optirnizat package pages pami path pattern pentland perspective perspectiveshape planning prados press problem proceedings programming quan recognition references results rosenfeld rouy sdpt semidefinite sethian shading shah shape shapefrom shum shupejrom siam simple sochen society software softwure solutions some source stochastic survey system tankus todd tourin tsai tutuncu under understanding uniqueness unrl using viscosity vision visionrind wellposed yeshurun zccv zeng zhang zheng
http://doi.ieeecomputersociety.org/10.1109/CVPR.2006.201	37	Multiple Face Model of Hybrid Fourier Feature for Large Face Image Set	academic accuracy achieved addison addition advance advanced advances advices algorithm algorithms also analysis angle approaches approximately atick augmented authentication banca bands bartlett based baseline basically because beijing belhumeur better biometric blackburn bone bowyer built bunch casasent challenge chang china class classes cognitive compared comparisons compensated component componentbased components computation computer computing concatenated conclusion conditions conf conference consists contributions corefaces correlation cvpr database demonstrated described description descriptor determination diego differences digital discriminant discriminate discrimination discriminative domain domains dordrecht each efforts eigenface eigenfaces eigenphases elastic evaluated evaluation experiment experiments extracted face facial fairly feature features fellous feret filter fisherfaces flynn fourier framework frequency frgc from frvt gain gender general generality gesture global gonzalez gradient grand graph grother group haibing hamouz heisele hespanha hoffman http huang hwang iapr ieee illumination image imaginary improvement insensitive integral intelligence intensity inter international into invariant joint journal kamei keen khosla kittler kluwer kostin kriegman kruger kumar lack learning lecture linear local machine machines malsburg marques matching maximum members merged messer method methodology micheals models modes moon more moreover mpeg multi multiple network networks neural neuroscience normalized notes novel object obtain other overview passionate pattern penev pentland performance performances person phase philips phillips poggio power press proc proceedings processing projected projection proper proposed protocols prove publishers random rate rauss real recognition references report representaion representation respectively results retrieval rizvi robust sadeghi sampling samsung savvides scheme science scruggs shanghai shift shows significant single society specific spectrum standardisation standardization statistical subspace successfully support system systems tabassi table tang technology test than thank that their theory there they three tolerant trans trolled turk uncontrolled understanding unsupervised using vector vendor verification versus vijaya vision wang washington wesley wiskott with within without woods worek workshop xiangsheng yamada yuan
http://doi.ieeecomputersociety.org/10.1109/CVPR.2006.68	112	Beyond Bags of Features: Spatial Pyramid Matching for Recognizing Natural Scene Categories	about access accomplishment achieves acknowledgments adaptable advantage agreement allocation alone alpes also analysis appearance approach approaches arregui aspects auer ballard based bayesian berg biases biologically blei boosting caltech caputo categories categorizers categorizing category class classes classifcation classification cnrs collaboration collected color computational computer conference consistently constructing content context correspondence correspondences cortex criminative crowley csurka cues cvpr dance darrell data databases datasets depth descriptors designed despite detection develop dirichlet discovering discriminative distortion diverse doorn eccv efros either entropy envelope evaluating even evidence examples explicit fact features feedback fergus field files foundation framework freeman frequency from full fussenegger gatica generative generic given glance global gool gorkani graf grants grauman graz grossberg hadjidemetriou hierarchical highly histograms hofmann holistic http hypotheses iapr iccv icpr ieee ijcv image images important improvement incremental indexing indirect indoor information inria inspirations international invariant inverted jordan journal kernel kernels koenderink large larger latent lazebnik learning local locally location lowe machine maire malik marszalek match matching maximum method methods model modeling models modules monay more motivated muller multidimensional multiresolution murphy national natural nayar nearest neighbor newly nine nition object objects odobez oliva opelt orderless orientation outdoor outperform over pages pami part partially parts pattern perona photos picard pinz place ponce power presence present probabilistic proc proceedings provide pyramid quelhas query raki receptive recipe recog recognition references relations relevance report representation research results retrieval rhone rubin russell scale scandinavian scene scenes schiele schmid science semantic sets shape simplicity sivic society sophisticated sorting spatial squire stand statistics still structure study such sults supported surprising swain system systems szummer take technical tested text texture that their they this three tools torralba towards training trans trivial tugraz tuytelaars ubiquitous uiuc under underscore unsupervised useful using variable video vision visual volume wallraven weak weights well willamowski with within without works workshop zhang zisserman
http://doi.ieeecomputersociety.org/10.1109/CVPR.2006.280	144	Simultaneous Registration and Modeling of Deformable Shapes	accuracy acknowledgments active adaptive agreement algorithm also analysis another appearance application applied approach arise associated banning based bases because behrenbruch benefit berge bias biased biermann border borders bregler british carroll central certain chai chichester choice closed comaniciu combinations come computer conf conference constraints constructs cooper cootes corporate coupled criterion currently cvpr data davatzikos deformable deformation deformations develop different dilation direct distorted dryden echocardiography efficiency either enforced epson equivalently essentially european exists exploring extend factorization first fitting focus form former formulation from function fusion gaussian generalized geometric goodall gower graham grant group have having health hertzmann however ieee image imaging importance improve individual information institute intelligence into jacob jing journal kanade kelion koschat krishnan latter left like linear lingoes lissitz loss machine mardia matrices matrix measured measurement medical mental method methods might missing model modeling models more motion myocardial national nemann noble noise nonrigid ones only orthogonal orthonormality outliers partly pattern point previous problem problems proc proceedings procrustes properly propose psychometrika quantifying real recognition recovering recovery references regard regional registers registration require research result rigid robust robustness rotation royal sample scho separate shape shapes shen siemens simultaneous simultaneously single smart snakes society solution solutions space spie stabilities statist statistical step steps streams structure such supported swayne takeo taylor that their them they this thus time tracking training trans transformation transformations under understanding uniqueness using varies ventricular vision weight weighted weights were when which wiley with within work xiao zhou
http://doi.ieeecomputersociety.org/10.1109/CVPR.2006.95	88	CSIFT: A SIFT Descriptor with Color Invariant Characteristics Alaa E. Abdel-Hakim and Aly A. Farag Computer Vision and Image Processing Laboratory (CVIP) University of Louisville, Louisville, KY 40292,USA	abdel academic accurate acivs adelson advanced affine after although america amsterdam anal analysing analysis angular applied arbitrary areas augustseptember autonomous average ballard barnard based basic bayesian belgium better between bolle boomgaard bottom brainard british brown brussel burghouts burt cases categorization certain challenging chatterjee code color colored colour combining communications compact compared comput computer concepts conditions conference considered constancy constraints contrast contribution copenhagen corfu crowley csift cvpr data denmark depends descriptors detect detected detection detector different directions distinctive eccv edition eigen eighth enough estimation european evaluation farag features field figure figures finlayson formulae fourth freeman from funt fusion geerts general generalized geusebroek gevers good graphics gray graybased grayvalue greece groups hakim head healey higher histograms hough ieee illumi illuminanation illumination image imaged images indexing information intell intelligence intelligent intensely intensities interest international invariance invariant invariants january john journal july june kanade keypoints kluwer kubelka laplacian large larger leads least lennie library lightscattering lindeberg local london lowe mach machine martin matched matches matching materials mechanisms methods middle mikolajczyk mohr more much multidimensional navigation nayar norwell note number object objects obtained ohta optical optics original pages part pattern patterns percentage performance philadelphia pictoseek pixels point pose proceedings processing publishers pyramid quantitative ratio receptive recognition recommended references reflectance region regions rejecting rejection repeatability repeated representation results retrieval road sakai scale scales schiele schmid science second segmentation settings shape show shown shows sift signs slater smeulders society some sons sources space specific springer stability statistics stiles still structures swain systems table tail than theory this those threshold thresholding thresholds times tion tool total trans transactions transform under using values varying verlag versus vision volume when which whose wiley wyszecki zmura
http://doi.ieeecomputersociety.org/10.1109/CVPR.2006.16	123	3D Reconstruction of Background and Objects Moving on Ground Plane Viewed from a Moving Camera Chang Yuan and Gérard Medioni Institute of Robotics and Intelligent Systems University of Southern California, Los Angeles, CA 90089, USA {cyuan, medioni}@usc.edu	acknowledgments across activity added adjustment advanced against algorithms also another approach approximating assuming background both bundle camera cannot computation computer conference configuration constraint constraints contract contribution contributions cvpr degenerate dense development directions discover effectiveness encouraging estimating exists experimental extended facilitate feature finding first fitzgibbon following formulated frames from funded future government ground hartley http iccv ieee ijcv initialized into investigating least lines manually mclauchlan method modern more motion moving multibody multiple object objects pages parametrization part patches pattern performs photomodeler planar plane point possible practical presence presented press problem proceedings prove recognition reconstruction references relations research results robust robustness sastry scale scenes segmentation sequences several soatto society solved static structure sufficient surface synthesis system tasks test there this torr tracked translation triggs under uniquely unknown vidal video view vision which will workshop zisserman
http://doi.ieeecomputersociety.org/10.1109/CVPR.2006.321	47	Unsupervised Discovery of Action Classes Yang Wang, Hao Jiang, Mark S. Drew, Ze-Nian Li and Greg Mori School of Computing Science Simon Fraser University Burnaby, BC, Canada {ywang12,hjiangb,mark,li,mori}@cs.sfu.ca	abnormality above academic accuracy accurately achieve acknowledgement across action actions activity adaptive against algorithm analysis another appearance applications april arms automatically averaged baseball based basketball behaviour believe belongie berg between blocked bobick body boiman boyd camel carlsson categories challenging chance chosen claim classes close cluster clustering clusters collection collections comparisons comput computational computationally computer conclusion conf conference context contexts corresponds could cutler cuts cvpr dataset datasets davis defender definitely deformable degrades demonstrated detecting detection difficult discovering down drew drives each edwards effective efficient efros emmcvpr employ energy essa european event events examples expensive experiment experiments exploiting expression extract face faces facial fast fergus figure forsyth freeman from further gait gavrila goes gong google grant handed have hayes hierarchical high hoey however human iccv ieee image images imaging intelligence interesting into invariant irani irregularities jain jiang journal kluwer label labeling labelling large laypast learned learning left linear little lncs location lower machine made maire making malik manual matches matching means method methods miller minimization moore mori motion move movement naive names nelson news normalized nserc object objects observers pages pairwise pami paper past pattern people performance periodic perona philomin picture pitcher player polana poses possible potential presented problem proc proceedings profiling programming progress pruning publishers puzicha quality rates real reasonable recogn recognition recognizing references repetitive representation requires results retrieval rgpin right rigid robust russell scheme search segmentation series shah shape shot significantly similar single sivic skates skating smart society some spectral speed spin still such suggests sullivan summarizing supported tasks technique template templates temporal that their these they this three throws time tracking trans transactions trials unsupervised unusual using vehicles video videre view vision visontai volume which white with without work workshop xiang yilmaz zhong zisserman
http://doi.ieeecomputersociety.org/10.1109/CVPR.2006.144	154	Image Matching Using Photometric Information Michael Kolomenkin Electrical Engineering Department,Technion Haifa 32000, ISRAEL	above accurate addition affine algorithm also analysis applications automated balanced barnard based baseline beardsley better bmvc bolles both cacm calculate cartography case causing changes chosen christoudias chum cipolla clearly color column columns computer computing conditions conference conform consensus consider constancy contours contribution corner correlated correspondence cumulative cvpr daylight deformations density deriche detected different differential distinctive distribution dolls drew easily eccv effective efficient electric enough epipolar estimation evaluation even every example explanation exploitation exploration faster features february figure finlayson first fischler fitting flusser frames fraser from function functions fundamental further geometry georgescu given good gool goshen gouet grayvalue grey have high histogram histograms hordley house huange iccv icpr identical ieee ijcv illumination image images important improvement improving inlier inliers insufficient integral invariant invariants josa judd july june keypoints laboratories large larger level lewd local loupiasc lowe macadam making many marked matas matches matching matrix meer method methods mitsubishi model mohr montesinos more moreover number numbers obdrzhalek observable observed obtains only original outliers pages pair pami paradigm pattern peak percent point practical presented probabilities probability proceedings progressive prosac provides quality random rate recognition references regions registration rejection removing report represent required research resized results retrieval salient same sample scale scenes schaffalitzky schmid school sculpture search sebea second seem segment segments separated separation several shadows shimshoni shown shows sift significantly simon sinclairy smith society some spectral steps stereo still subsequent successful such survey synergism table taken technical techniques temperature tested text texture than that there thesis they this thus tianb time true tuytelaars twice typical under university updated urban used value values versa vice viewpoint views vision were when which while wide widely window with wood wyszecki zero zisserman zitova
http://doi.ieeecomputersociety.org/10.1109/CVPR.2006.78	140	Combined Depth and Outlier Estimation in Multi-View Stereo Christoph Strecha, Rik Fransens and Luc Van Gool K.U.Leuven-ESAT-Psi Kasteelpark 1, 3001 Leuven, Belgium	account algorithm algorithms analysis appear applications approach approximate baseline bayesian belief bing binocular cambridge camera carving chai coherent comaniciu comparing complete computer conference cooperative correspondence cuts cvpr data dempster dense depth detection dpvt dynamic eccv estimation evaluation faugeras feature features field frame francisco fransens freeman from gargallo geiger generalizations gool graph handling heung hinton ieee ijcv image images incomplete incremental inference intelligent jian justifies kanade kang kaufmann keriven kolmogorov kutulakos ladendorf laird level likelihood machine mapestimation maps matching maximum mean meer method methods modeling modelling morgan mrfs multi multiframe multiple neal networks ning occlusion occlusions opper other outliers page pages pami parallel pattern pearl plausible pons presence preserves press probabilistic problems proccedings proceedings produces propagation publishers reasoning recognition reconstruction references reflections registering robust rubin saad scene scenes scharstein seitz sequences shape shift sing soatto society space sparse spatially specular statist stereo stereovision strecha sturm symmetric systems szeliski taxonomy that theory this toward understanding using variants variational view views vision weiss wide yedidia yeung yezzi yuille zabih zitnick
http://doi.ieeecomputersociety.org/10.1109/CVPR.2006.41	105	A Simple Bayesian Framework for Content-Based Image Retrieval	abstract acknowledgements active adjust advances advantages aerial alexei also analysis animal annotation applied ashley assfalg automated automatic based bayesian behold belong bernardo bimbo binarizing binary bircol birtex boat browsing building butterfly calo castle category cavern cbir cell choice church civr classification clouds coast college color combined combining comparison compression computed computer conclusions conference consisting contains content continuum cord corresponding criterion cvdb cvpr cybernetics darpa data databases dataset density described desert developing different directly distance door drawing early easily efficiently eiffel either error estimation evaluation example examples experience experiments explore expo extend extensions extract extraction factor feature features feedback fields find finds fireworks first flickner flower fractal frames framework free from fruit functions future gaussian ghahramani global good gorkani gosselin group grouse gupta hafner harmonic have heesch heller help here histogram house howarth however http huang hyperparameters icme icml ieee image images imperial implementation implicitly incorporate incorporating index indexing informalondon information initial input instead intelligence interesting international intersection into jain kitchen labeled labelled lafferty lastly learning lights like likelihoods likely lippman look machine manually many marginal mars matrix maximizing meets mehrotra method methods miller minimize minimum minka model mori most mountain mountains multimedia multiple multiplication negative neural niblack nnall nnmean nonparametric obtain only other over pages pala papathornas parameters partly pattern penguins people percentile perception perform performance person petkovic pichunter pickering plain plan positive preprocessing principles probability proceedings processing project psychophysical qbic queries query querying recognition references relevance reptile respectively results retrieval revised river robust ruger santini sawhney scaling schofield seaches search searching searchvis seemed semi sets setting show sign signal since single smeulders smith snow society some space sparse specific specifies stairs steele subsequent such sunset supervised supported system systems table taking tamura target text textual texture textures thanks that then theory there these this threshold tion tool tower train tran trans transactions trecvid trees trieval turtle type types union unlabeled urban used user uses using values varies vasconcelos vector vectors video vision visual volcano volume water waterfall well when where which while white wiley with woman word words work workshop worring would yamawaki yanker yavclinsky yavlinsky years yianilos zebra
http://doi.ieeecomputersociety.org/10.1109/CVPR.2006.325	104	Using Language to Drive the Perceptual Grouping of Local Image Features	access analysis annotated annotation approach barnard based berkeley bernoulli blei brown california carbonetto carniero cascia choice class combining communication computational computer conference content contentbased contextual cues cvpr data dickinson distinctive division duygulu eccv effects estimation evolution feature features feng fergus fixed flexible formulating forsyth frameword freitas from gabbur general google grounded grouping guru hoogs iccv ieee ijcv image images invariant jepson jordan june keypoints knowledgebase language large lavrenko learning lexicon libraries linguistic linguistics local lowe machine manmatha matching mathematics meaning mercer model modeling models multiple naacl natural object pages parameter pattern perona pictures pietra problem proc proceedings recognition references relevance report retrieval rittscher scale schmiederer science sclaroff segmentation semantic semantics sethi shape sivic society spatial spoken statistical stevenson stien structured supervised syntax technical text textannotated textual towards translation university unsupervised using vasconcelos video videos vision visual visually vocabulary volume wachsmuth wide word words workshop world zisserman
http://doi.ieeecomputersociety.org/10.1109/CVPR.2006.298	86	Supervised Learning of Edges and Object Boundaries	acknowledgements action active additive alfred algorithms amodal analogies application approach authors award basri bayes bayesian belongie boosting boundaries brace branson brightness canny career carlsson cfda characterization chen classification clustering color comaniciu combinations company completion complexity computation computational computer conclusions conditional conference confusion contour contours corporate coughlan cues curless cvpr decision detect detection different discriminative discussion done dorin dubuc eccv edge engaged evaluating expressed feedback fellowship fields findings fowlkes francisco freund friedman from geisler geman generalization generate gestalt grant grouping harcourt hastie hertzmann high human iccv icml ieee igert ijcv illusory image images invariant itly jacobs jedynak jones knowledge koffka konishi kriegman learning level like line linear local logistic make malik many marr martin material mathematical modal model models motion mouse multiple must natural naval necessarily neural object office oliver opinions pami parsing pattern perceptual performance primal primarily principles probabilistic proceedings psychological psychology ramesh random real recognition recognizing recommendations references reflect regression report represent research roads robust salesin salience samples satellite scale scenes schapire science sctv segmentation segments shape shapes siemens siggraph singh sketch sketchability sloan society stanford statistical statistics stochastic structure sullivan support supported taylor tech testing texture thank theoretic theory thesis this those tibshirani time tracking tree ullman understanding unifying university using valuable view views viola vision visvanathan washington were while williams with without work would york yuille zucker
http://doi.ieeecomputersociety.org/10.1109/CVPR.2006.10	125	3-D Shape Reconstruction of Retinal Fundus Tae Eun Choe1, Isaac Cohen2, Gérard Medioni3 Institute for Robotics and Intelligent Systems University of Southern California, {tchoe, icohen, medioni}@usc.edu Abstract	abramoff accurate accurately acknowledgements algorithm algorithms alignment also analysis anandan applications approach april area areas atsumi authors automatic bartoli based been belief biomedical boundary calif cambridge cases challenging change choe chum code cohen color complex computer computing conclusion conf conference contents correspondence criteria cuts cvpr darpa data database december defence defined dense depth detection dick difficult dimensional disc displays doheny dominant edge effects efficient effort energy epipolar estimated estimating estimation evaluate evaluation exception express faculty faez fall fluorescein focused fovea frame from fully fund fundamental fundus future geometry germany global gluckman grant graph grey hanna hartley health higgins hirschm human iccv icpr ieee ijcv image images information innovation institute institutes international ismda jackson james jiang journal july june kauai kikinis klein kolmogorov kumar landmarks lesions levels like ller longuet magarely march matas matching matrix maximization medical method middlebury midgraph minimal minimization minimize mircea moallem modal modeling monterey mordohai morphological most multi multimodal multiple multiresolution mutual nakajima national nature nayar necessarily nicolescu nolinear november occlusions official optic pair pairs pami paper parallax parameters partially participating pattern performance philippos plane podoleanu point press proceeding proceedings processing projections propagates propagation proposed recognition reconstructed reconstructing reconstruction reconstuction recovery rectifying reduction references registration reliable represent resampling research responsibility retina robust rogers scene scharstein search segmentation semi sept september sequence sequences several shape shum since skin society solely source space stereo sturm stuttgart submitted supported surface szeliski task taxonomy techniques tested thank that these this three through trans transformations tree unaffected under university using vascular vessel vessels view views viola vision visual visualization volume walter wavelet webpage well wells werner with work workshop would yfeatures zabih zheng zisserman zumberge
http://doi.ieeecomputersociety.org/10.1109/CVPR.2006.70	136	Binocular Stereo Dense Matching in the Presence of Specular Reflections Wei Zhou and Chandra Kambhamettu Video/Image Modeling and Synthesis (VIMS) Lab University of Delaware, Newark DE 19716, USA wzhou,chandra@cis.udel.edu	accurate added address advances algorithm algorithms allows america analysis anandan applications approximation assumption attempt avoid ball barn based baseline bdisc because belhumeur beyond bhat binocular blake boykov brdf brelstaff breneman brown bspe buntex burschka chantler color compact comparing components computational computer computing conf conference constancy constraints contd contribution correspondence corresponding cuts cvpr cycle data davis demonstrated dense depth detecting developed development devised difference diffuse diffusespecular drbohlav eccv effects employment energy estimating european evaluation experiment experiments fail fast fields figure filter find frame framework from further future geiger general global gradient graph hager ibrate iccv icpr ieee ijcv image images including increase intelligence intl invariant investigate journal kalman kanade kang kaufmann kriegman ladendorf lambert lambertian left light local machine major mallick matching matthies measurement method metric middilebury minimization minimum model modeling monterey most motion multi multibaseline multilevel multiple nayar novel november occlusions okutomi optics other pami parametric pattern perform photometric pixels points poster presence previous problem problems proc proceedings publishers rate ratio real recognition reconstructing recovery refelction references reflection reflections rendered research results right robust robustness roughened saad sawtooth scharstein schulte segmentation separate separation sequences sets shafer shum soatto society sparrow specular stereo surfaces systems szeliski table taxonomy tend their theory this three tong torrance tpami traditional trans transforms transport understanding unlike used using utilizing veksler video view violated vision visual volume wang when which will windows with woodfill work works workshop yang yezzi yuille zabih zickler
http://doi.ieeecomputersociety.org/10.1109/CVPR.2006.300	120	Surface Reconstruction of Bone from X-ray Images and Point Distribution Model Incorporating a Novel Method for 2D-3D Correspondence	accuracy accv active adding advantages aided algorithm anal anatomical another apparent application approach arthroplasty assisted automatic based been benameur berlin besl best between biomech biomed biplanar blanz bone bones both bousson bras brechbuehler canny cases certain chui closed colchester comp comput computational computer computeraided conclusions conference contours control cooper cootes correspondence cortical cross ctfree cvpr davies deformation dense description detected detection different difficult digital dimensional direct discussions distal distribution doing dose duncan eccv edge engin enhanced enough exclusion extract faces femur find first fitting fixation fleute fluoroscopy fomr fracas fraction fracture framework from gerig good graham grap guid guided guise have hertzmann higher hofstetter hybrid ieee illustrating imag image images imaging implants important improve include information injective instead ipmi isbi iterative iteratively joshi joskowicz kuebler langlotz laporte lavall lecture length level long lowe mapping marx matched matching mckay means medical method methods miccai mignotte milgrom model models more morphable morphing morphologies much narayanan navigation nayar nearest need neighbor nolte nonrigid notes novel object only operator optimal optimization orthopaedic outliers overall pairs paper parameterization parameterized parent part pattern pixels plate point points possible preliminary presented preservation proceedings process processing projections property proposed proximal quite rajamani rambo rangarajan recognition reconstructed reconstruction references registration respect rigid robust robustness sati schultz science scoliotic segmentation shape shapes shum siggraph simkin skalli slomczykowski smooth society some space sparse specially spinal splines springer stage statistical step stereoradiography still structures study styner successfully such surface surfaces surg surgery surgical symmetric synthesis system target taylor that their there thin this three topology total training twining under uses using validated vertebrae vetter visarius vision visualization which with zheng zorin
http://doi.ieeecomputersociety.org/10.1109/CVPR.2006.86	109	Context and Hierarchy in a Probabilistic Image Model	abstract academic accurate acknowledgement advances amit analysis anderson annals applications applied approaches architecture argue artificial artificially aspects assessment asymptotic background barlow based bayesian been berlin between bias biederman bienenstock black blanchard boosting borenstein bouman brain brief brown building cambridge capable careful carmichael cerebral character chen chicago choice chomsky class classification coarse cognition cognitive comp complete complexity components composition compositional compositionality computation computational computer computing conf conference connectionism consistency consistent control corporation cortex cortical counter crandall critical cross cvpr data davis deal define density dependences descriptions designs detection detections deterministic dilemma directional discussion distinction division doursat downsegmentation duda dynamic earlier eccv editor editors effective efficient either emory empirical encoding equip eric essen essence esssai estimation experiments experts false familiar fast features feilds felleman felzenszwalb fields filtering fine fleming flow fodor forcing foregrounds formulation foundations framework francisco freeman freund from fukushima full function geman generalized ghahramani gibbsian goal grammars great grenander hand harper hart have hierarchal hierarchical hinton holden holland hopkins human huttenlocher iccv ieee image images include including inference inferior information integrated intelligence interest intermediate interpretation interpretations intl invariant johns johnson jones jordan jour journal july khudanpur koch krempp krishnaiah labeled language laplace large learning lehmann levels license limit linguistics local locations logothetis lowe lowlevel machine made manbeck math mathematical mathematics mcclure mceneaney mentioned method methods mode model modeling models mozer multi multiclass multiple murphy namely naquet natural nature neocognitron neocortex network networks neural neuronal neurons neuroscience nips nonlinear nonparametrics north noticing object objects observations occur ostendorf other pages pami papers parse parsing part parts parzen pasztor patchwork pattern performance perona perspective petsche philosophique plate poggio pollak position potter predictors predominantly president press primate priors probabilites probability problems proc procedures proceedings processing proper psychological pylyshyn qualitative quarterly random rank ranks reading real reasoning recognition references regarding report representations result reusable review riesenhuber rissanen robust role rooted rosenfeld roth sagerer sali same scale scene scenes schapire science search seeking segmentation sequence sequential shape share sharing sheinberg sherman signals siskind smoothing socher society solving some spatial specific speech springer stat state statist statistical statistics stochastic stone storkey strategy streams strongly structure stuff subgraph submitted subtrees suggest supp supplying survey syntactic syntax system systems technical temporal testing thank that their them theories theory these this three time torralba trans transactions translation tree trees trouve truscott ullman understanding unifying university using validation validatory vapnik variance verlag vidal viola visics vision visual volume wahba what whereby wiley williams wish with without world would york
http://doi.ieeecomputersociety.org/10.1109/CVPR.2006.74	77	Bottom-Up & Top-down Object Detection using Primal Sketch Features and Graphical Models	achieved affine agarwal algorithms allowing also amit appearance approach approaches arbitrary aspects attribute attributed automatic babu based baseline been belief bernstein between bhatia black blobs bmvc boosted both bottom burl cascade categories categorization category cgip chum class classification codebook cognition combination combined compact complex computational computations computer conclusions conference constant construction corners could cows crandall currently curves cvpr description descriptors detection detectors distinctive down eccv edge edges efficient enabled estimation excellent exhaustive expectation exploitation extending extraction extremal faces facilitate feature features felzenszwalb fergus filling finding form forsyth framework freeman from general geometric geometrical geometry global good grammar graph graphical grayvalue hall have higher hoffman horses huttenlocher iccv ieee ihler ijcv image implicit improved improving incorporated incorporation information interest interested introduce invariant invariants ioffe isard iteratively jones july junctions keeping keypoints kimia kokkinos kumar learned learning leibe leonardis level like likelihood limbed lindeberg linear local localization loose lowe maragos matas matching mathematical maximally maximization method methods mikolajczyk model models modern mohr more multi nevatia nonparametric object objects pajdla pami parsing part parts pattern patterns people performance perona photometry pictorial point ponce poses prentice primal primitive primitives priors probabilistic proceedings processes profiles propagation provided pursuing rapid recognition references refine regions related representation rest resulting results retrieval richards ridge ridges robust roth same scale schaffalitzky schiele schmid segmentation selection sets shape siddiqi sigal simple sketch sketchability slcv small society sparse spatial stable statistical stereo straight streams structure structures sudderth synergy syntactical systematically templates than their theory this thus torr tracking training trans unordered unsupervised urban used using view viola vision visual weber welling where while wide willsky with without work workshop zisserman
http://doi.ieeecomputersociety.org/10.1109/CVPR.2006.90	55	Correlated Label Propagation with Application to Multi-label Learning Feng Kang1 Rong Jin1 Rahul Sukthankar2	aaai academic additive algorithms alon annotation applied approaches approximate aspects athitsos barnard based bayes binary boostexter boosting cascade categorization category cbsa chang cikm circuits class classification computation computer conceptual conference content contentbased costly crammer cvpr data description directed discrete distribution document duygulu dynamical eccv effect efficient entropy equation equations evaluation family feedback fields finally fixed forsyth freitas functions gaussian ghahramani gong graphs harmonic hierarchical hofmann hyuk icml ieee image indexing induction inequality infinite information informed into issue jair kopf label labelled lafferty language latent learning lexicon machine machines manjunath margin maximum measures method methods mixture model models multi multilabeled multimedia multimodal nearest neighbor networks neural nips object obtain online optimization outputs parametric parker pattern point preference press proc proceedings processes provost ranking recognition references relevance retrieval rousu saito saunders schapire schol sclaroff search semantic semi shawe sigir similarity singer smoothing society soft special statistical structured study substituting sung supervised support sychay systemfor systems szedmak taylor tech tesic text trained training tran trans translation tree tresp ture ueda using vector video vision vocabulary weiss when williams with workshop yang zhai zhou
http://doi.ieeecomputersociety.org/10.1109/CVPR.2006.278	94	Shape Topics: A Compact Representation and New Algorithms for 3D Partial Shape Retrieval Yi Liu1, Hongbin Zha1, Hong Qin2 National Laboratory on Machine Perception, Peking University1 Department of Computer Science, State University of New York at Stony Brook2 {liuyi, zha}@cis.pku.edu.cn, qin@cs.sunysb.edu	acad acknowledgements algorithms allocation also analysis applications approach approaches asymmetric asymmetrical automatic basis bayesian belongie benchmark between blei blue both buckley bulow cambridge carlo categories category chazelle chen china choose circles class classes classification cluttered coarse collections compact comparable comparison computer computing conclusion conference contexts correct correspondence corresponds cross cviu cvpr data decided demonstrated descriptors diedrich difference different dirichlet discovering dissimilarity distinctive distribution distributions divergence dobkin does each eccv effective effectiveness efficient efros eleven engine establish example experiments extensive false features fergus figure file finally finding fingers first form framework freeman from frome function funkhouser generate geometry global good google granularity graphics green griffiths halderman hand hard have hebert hierarchical high histogram however huber iccv identical ieee ijcv image images included infer information intentionally interested international introduced investigate irregular jacobs january johnson jordan journal kazhdan kernel keypoints kiefer kolluri kumar label latent leads learning list lowe machine maeder make makes malik management mapping margin mark matches matching matei measure mesh message method model modeling models monte mori most multi multiple natl natural nearest nkbrp note nsfc object objects obtain obtained october only optimal osada palm pami paper parameters part partial parts passing pattern perhaps perona planitz point points previous princeton probabilistic problem proc proceedings processing projection proposed puzicha quantization query radial range ranking rate recognition recognizing references referred regional remove removed represented research results retrieval retrieved robust rolf rusinkiewicz russell salton same sample sawhney scaleinvariant scene scenes scientific search second select selected semantic seven shan shape shapeme shapes shilane show shown siggraph signature similar similarity sivic society spin statistics steyvers supported surface svms table term terms tessellation testing text that there therefore these thesis this topics train training trans underestimated university used using utility validate validation variational vector vision weighting which williams winn with words work wrist yellow zisserman
http://doi.ieeecomputersociety.org/10.1109/CVPR.2006.193	38	Multi-Aspect Detection of Articulated Objects Edgar Seemann, Bastian Leibe and Bernt Schiele Computer Science Department Darmstadt University of Technology	adaptive appearance applied assembly background based bayesian belongie between bhattacharyya bmvc bull calcutta cars categorization combination components computer conference contexts crowded cvpr dagm dalal defi descriptors detecting detection detectors distance distributions divergence eccv edgelet evaluation example faces feature features freeman gavrila germany gradients grimson hierarchical histograms human humans iccv icpr ieee ijcv image images invariant jones kanade leibe lncs local malik matching math mean measure method mikolajczyk mixture models mohan motion multi multiclass multiple multiview murphy nevatia object occluded oriented pages pami papageorgiou part partially pattern patterns pedestrian pedestrians performance poggio populations probabilistic probability proceedings puzicha realtime recognition references robust scale scaleadaptive scenes schiele schmid schneiderman search seemann shape sharing shift single snow society springer statistical stauffer system template their torralba tpami tracking trainable transforms triggs tuebingen using viola vision visual volume zisserman
http://doi.ieeecomputersociety.org/10.1109/CVPR.2006.269	21	Jim Chan	able actually added addition after against allowed alone also american amin analysis apart apparent appearances approach arabic automated average back barely based because been before beirut better bigrams bock brevity build built bulbul burges cantly carefully cation center character characters classifi common comparable compare compares complex component composed comprehensive computer conference connected containing contents converted correct could cursive curve curves cvpr data datasets decision demo denker described description determining different discriminate discrimination discriminative discussion document documents duda earlier easy edwards eigenvalue elongated engineered enitdatabase enough evaluation even except explanation extension external figure form forms forsyth fortunately found frequent from further future ghmm given giving groups handwriting handwritten hart have here historical holistic however http hull icdar ieee ijprai importance imposes impossible improved include increasing index individual information itself jafet just kanungo kernel kitab knowing language large larger last latin lavrenko learning lecun left letter letters lexicon ligature ligatures likelihood likely line lines list made maergner maire making manmatha manual manually manuscripts many marked marton method methods mistakes model modeled models modern more most much muller need needed networks neural news nips nohl nonlinear note notice occurrences offline online only ordering other overlaid overlap page pages pattern pechwitz performance plamondon postal precision probability problem proceedings process products projects provide quite qwef rate rath readily reasonable recall recognising recognition references removal report reported represents results retrieve returned right robinson roman same scene scholk search searchable searched searches searching senior sequence shape shotgun shows signifi since smola snowball society some spacing split srihari stacked stand strong stronger structural studies summary survey system technical techniques template templates tended term texts that their thick thin this though three times total tpami train training transcribe transcription transition transitions tree trie turned unique university used useful using variant vesom visible vision volume washington well were which wiley will with without word wordlist words work would york zero
http://doi.ieeecomputersociety.org/10.1109/CVPR.2006.27	93	A Generative-Discriminative Hybrid Method for Multi-View Object Detection	academic alternative appear appearance applications approach approaches approximate artificial arts attributed auer austria automatic automatically barrow based belief berkeley between binary boosting brady categories chap class columbia combinatorics comparison component computer conclusion conference confirmed crandall cvpr cycles data derive description descriptions detection develop discovery discriminative division docs effective efficient erdos experiments exploring felzenszwalb fergus framework freeman functions fundamental fussenegger generalizations generative generic graduate graph graphs graz have hertz hillel holub http hybrid iccv ieee image incorporate inference intelligence international into intuitive invariant jones jordan journal kadir labeling learning learns likelihood loopy machine manner methods millennium model modeling models multi networks object opelt optimization outperforming pages part partbased partition pattern performance perona picture popplestone power presented press previous priors probabilistic proceedings processing propagation pure random ratio real recognition references region relational relations report rigorous robust robustness saliency scale scheme school sciences seattle semidefinite similarity society sourceforge spatial spencer stable statistical structure table technical then theory thesis time towards uncertainty understanding university unsupervised using variations view viola vision wainwright washington weber weinshall well welling with work yedidia york zhang zisserman
http://doi.ieeecomputersociety.org/10.1109/CVPR.2006.79	66	Combining Cues: Shape from Shading and Texture	analysis angles anisotropic another back based behavior berkeley better bmvc bottom boundaries calibration camera case chen choe cloth comparison compute computed computer conf conference constraints cryer cvpr david deformation deforming dense describe described detection dynamic eccv eecs element fields figure film finite first forsyth frames from frontal geometry goldgof graphics grid hartley homogeneous horn huang ieee image integrability interface interpretation irradiance jects julien kashyap lepetit lobay local localizable maps mass method mirror model models motion nonrigid nonstationary object obtaining obtechnical only opaque orthographic pami pascal pattern perona perspective pilet press proc proceedings projections provide provot quality real recognition recover recovered recovering references refinement reflection reflections region regular rendering report result retexturing rich rigid ryan sarkar savarese scene second sequence shaded shading shah shape shirt single smooth society spring structure subsequently surface survey texton textural texture textured that this time total tsai tsap ullman using verify view views vincent vision visual visually volume white without xavier yoonsik zhang zisserman
http://doi.ieeecomputersociety.org/10.1109/CVPR.2006.287	69	Spatial Reflectance Recovery under Complex Illumination from Sparse Images Li Shen Haruo Takemura Osaka University Osaka 560-0043, Japan	acquiring analysis analytic annual applied approximation arvo based basri boivin brand brdf complex computer concept conference construction current cvpr dana datadriven debevec diffuse dimensional duiker effectively eurographics face field framework from functions gagalowicz gallery generally ginneken global glossy goesele graph graphics greenberg hanrahan hawkins heidrich http human iccv ieee ikeuchi illumination image images intelligence interactive international inverse jacobs january kautz koenderink lafortune lambertian lemmo lensch light lighting linear machine malik many marschner materials matusik mcmillan measurement might model modeling models more nayar nonlinear object observation optics pattern pfister photographs predicting probe probes proceeding proceedings processes processing ramamoorthi real recognition reconstruction recover recovering references reflectance reflection removing rendering result robustly sarokin sato scenes seidel shape siggraph signal single society spatial spatially specular speed subspaces surfaces tchou techniques terms texture theoretical torrance transactions variability varying vision westin wheeler which workshop world
http://doi.ieeecomputersociety.org/10.1109/CVPR.2006.245	71	Reconstruction in the round using photometric normals and silhouettes.	accuracy achieve achieves algorithm allows also ambiguity anal angle apart apparent applications assume automated based been belhumeur better between blake bolles brdf brdfs cacm calibrate calibrating calibration camera capture cartography chandler cipolla circular claus combining computer concept conclusion conf conference consensus constant consuming contours contribution could cremers curless cvpr davis dbrohlav december deformation degrees demonstrated detailed deviation different direction drops each effect efficient efficiently eliminating epipolar error estimation eurographics europ evaluation every example extended extract fact favaro featureless figure first fischler fitting fitzgibbon flash found frame frames from frontier fully further fusion general geometry give goldman greater ground have hernandez hertzmann highlights hull ieee illumination illuminations illumniation image imising importance information inhomogeneous intell intensity intl investigation journal just knowledge known kriegman lambertian laurentini length light lighting mach main manual mean measured mendonc method model modeling motion multi necessary need nehab normals novel november number object objects pages paper paradigm particular passive paterson pattern photography photometric pixels points positions powerful practical precalibrated precise presence presented previously proc proceedings process profiles purpose qualitatively ramamoorthi random ransac recognition reconstruct reconstruction recover recovered recovery references reflectance relief robust runs rusinkiewicz same sample samples scheme schmitt segmentation seitz self sequence setup shape sharing shedding shown shows siggraph significance silhouette silhouettes single soatto society source spatially special specular standard stereo stereoscopic structure subsequences such sufficient surface technique techniques that theoretical this thus time total trans treuille truth under understanding uniform unifying unknown used using variety varying view viewable vision visual vogiatzis volume where which with wong yang yezzi ysis yuille zhang
http://doi.ieeecomputersociety.org/10.1109/CVPR.2006.302	78	The Bottleneck Geodesic: Computing Pixel Affinity Ido Omer Michael Werman The Hebrew University of Jerusalem, Israel Ross Building, Givat Ram Campus 91904 Jerusalem, Israel {idom,werman}@cs.huji.ac.il	acceptable achieve achieved adaptive advances affinities affinity algorithm alone although analysis annals appealing appearance application applications applied apply approaches attention based believe benchmark berkeley better between bottleneck boundary buhmann calculating chan chang cheng choice clustering color comaniciu combining comparable comput computationally computer conf conference connexion consuming continue cues cvpr data dataset demonstrated density detection different dijkstra discontinuities distance distances eccv effect enough estimation euclidean european exploit feature features figure fills fischer fowlkes framework function functionals functions further geodesic geodesics geometry global gradient graphs group grouping histogram http iccv ieee ijcnn image implementation important information input intend international jain jiang journal june kinds learning lecture leung level limited lines local malik many martin materials mathematik meer method minimizing model more multiphase mumford ncut note notes numerische octorber omer only optics optimized other oxford page pages pairwise park particular patchbased path pattern phasing polynomial problems proc proceedings projects prospects rattray received recognition recognizing reconstruction references relies representation representing results robots robust running sample sampling scene schowengerdt science segmentation segmentations segments september shah similarities similarity society spaces specific spectral spokoiny statistics such tasks tenth textons texture than that these this threedimensional time times traditional tuceryan university used uses using various vese vision visual visually volume wang washington werman where will window with
http://doi.ieeecomputersociety.org/10.1109/CVPR.2006.328	129	Vertical Parallax from Moving Shadows	accurate bombay bouguet chang chuang compositing computer conference correspondences criminisi curless cvpr dense desk diego efficient encoded france from goldman graphics iccv ieee images india institute international january july light london matting metrology multiple nice october pattern patterns perona photography procams proceedings recognition references report salesin shadow siggraph single society springer szeliski technical technology temporally transactions uncalibrated using verlag vision visual your
http://doi.ieeecomputersociety.org/10.1109/CVPR.2006.301	107	SVM-KNN: Discriminative Nearest Neighbor Classification for Visual Category Recognition Hao Zhang Alexander C. Berg Michael Maire Jitendra Malik Computer Science Division, EECS Department Univ. of California, Berkeley, CA 94720 {nhz,aberg,mmaire,malik}@eecs.berkeley.edu	adaptive algorithmic algorithms american anal analysis analytic appearance application applied approach association available backpropagation bags based bayesian belongie benchmark bengio berg berkeley beyond bickel blur body boosting boser bottou bromley buhmann california caltech cambridge categories categorization class classification classifiers code cognitive combining compositional comput computation computer conclusion conference configurations context contexts convex correctness correspondence cortex cover crammer cristianini csail cvpr dags dana darrell data deals denker department dependent detection difference dimensional discriminant discriminative dissimilarity distance distances distinctive distortion division document doksum domeniconi duin eccv edition efficient enable estimating estimation european evaluation examples excellent experiments features fergus fisher fize form forms found framework francisco freeman from functions generalized generative geometric ginneken goldmeier gradientbased graph grauman growth gunopulos haffner hall handwritten hastie have henderson holub howard hubbard human hybrid hyperplane iccv ieee image images implementation incremental information inspired intell international invariance invariant issues jackel journal june kaufmann kernel kernels keypoints koenderink large lazebnik learn learning lecun leung levina lncs local localized london lowe mach machines malik margin marlot match matching materials mathematical methodology microarray model models morgan mori multicategory multiclass murphy mutch natural naturally nature nayar nearest nearestneighbor neighbor network networks neural nips november object ommer paclik pages paper pattern pekalska perceived percentage perona platt poggio ponce prentice press problems procedures proceedings processing propagation proposed psychological psychology publishers puzicha pyramid radiance rate real recognition recognizing references reflectance regions report representing results retrieval rosch rule sackinger satellite scale scene schmid scholk science serre sets several shape shapes sharing shawe show simard similar similarity singer single society sparse sparseness spatial speed springer statistical statistics support surfaces system table tangent taylor technical template tested textons texture that theory thesis this thompson thorpe three tibshirani torralba trade training trans transformation trick tricks under university using vapnik variety varma vector verlag version victorri vincent virtually vision visual visually volume wahba wang welling wgmbv where which winter with wolf workshop world zhang zisserman
http://doi.ieeecomputersociety.org/10.1109/CVPR.2006.267	124	SDG Cut: 3D Reconstruction of Non-lambertian Objects	affected ahuja algorithm algorithms also appearance applied applying asian associated based belhumeur between bias biases boykov brdf brdfs camera cannot carving change cipolla color coloring comparison complex comput computer computing conference consistency constancy constraint constructed controllable conversion correspondence current cuts cvpr datasets davis dealing degree dense discretization distance dyer each eccv effectiveness energy estimate evaluation evolution example experimental exploiting face faugeras flexible flow fluctuating frame from further general generalized geodesics georghiades global graph handle helmholtz hertzmann highlights however iccv ieee image incorporating initial initialization intersection into invariant investigate iterations keriven kolmogorov kriegman kutulakos lambertian level light london measure mesh method methods minimal minimization minimum model more multi node novel obtain only optimal other page pages pami part pattern performance photo photometric photorealistic plan pollefeys possible principles problem problems proc proceedings processing progressive proposed radiance real reciprocity recognition reconstruction reduce references reflectance regions restricted result results robust scene scharstein scheme search segmentation seitz shading shape should show shown signed since smoothness soatto society space sparrow specified specular springer stereo stereopsis such surface surfaces szeliski tales taxonomy textureless than that theory these this topology torr torrance trajectory trans transform transforms transport uncalibrated under underst using variational varying verlag vertex view vision vogiatzis volume volumetric voxel wang washington welch with within worse yang yezzi zabih zickler
http://doi.ieeecomputersociety.org/10.1109/CVPR.2006.29	24	A Graph Based Approach for Naming Faces in News Photos Derya Ozkan and Pinar Duygulu Bilkent University, Department of Computer Engineering 06800, Ankara, Turkey	achieved acknowledgements across algorithm algorithms also appearances applied apply applying approach approx approximation associated association average baker barnard based before berg better between binary blei captions career charikar chellappa chen city civr close collections combinatorial compare compared component components computer computing conf conference consisting construct constructed convert correlating corresponding cvpr data dense densest describe described descriptors difficult discussion dissimilarity distances distinctive does dublin duygulu each editors edwards error evaluation everingham expressions face faces feature features find finding forsyth freitas from given grant graph greedy gross group handbook hauptmann have however ieee ignores illumination image images increase individuals information interest into ireland jain jordan journal kanade keypoints labeling large learning less literature local london lowe machine mainly malsburg manjunath matching matrix matthews measures memory method methods mikolajczk minutes modal multi name names neural news nips number numbers obtained optimization other over paper partition pattern pentium people performance perin person phillips photograph picture pictures points pose precision problem proceedings processing prone property propose queried querying realistic recall recognition references represent representations require requires research results retrieval rosenfeld samples satoh scaleinvariant schmid search second seek sets shot sift similar similarity since singapore sivic society solely some space specific spotting springer still structure study such summary superior supervised supported survey surveys system systems takes text than then this thus training transformed tubitak university used using values variations verlag very video violate vision visual weighed weighted which with words work workshop yang yield zhao zisserman
http://doi.ieeecomputersociety.org/10.1109/CVPR.2006.294	139	Stereo Vision in Structured Environments by Consistent Semi-Global Matching Heiko Hirschmulleš r Institute of Robotics and Mechatronics Oberpfaffenhofen German Aerospace Center (DLR), 82230 Wessling, Germany.	about academic accuracy accurate airborne algorithm algorithms alignment analysis appear approach apriljune area areas assumption august austria back background based because been belief best better between birchfield black bleyer borders boundaries calculated camera cases chairs chances change checking circle circles cleaner color column comaniciu common compared compares comparison compete complete complex complexity computer computing conclusion cone cones conference consistency consistent constraints consumed correct correlation correspondence corresponding counted csgm currently cuts cvpr dagm davis demonstrates dense depth despite details diego differences different differs discontinuities discontinuity disparities disparity does during edition efficient environment environments error errors especially estimation evaluation even examples execution extensions fail falls feature features figure filtered find fine first foreground frame from front furthermore fuzzy gelautz global graph ground handle handled handling happens have high hirarchical hirschmul hirzinger however hrsc huge hypothesis ieee image images improve improvements increases india individually indoor information intelligence intensity international interpolated interpolating interpolation into invalid isprs january journal june just kang kitchen kolmogorov labeled lack last layered left light like limitations lncs lower machine madison maintained make many maps mark marks matching maximization maximum mean meer method methods middlebury mismatches model more most much mumbai mutual near nister object objects obvious occluded occlusion occlusions office online only other pages part pattern percent perform performance performed performing performs photogrammetry pixel pixels pixelto place places planar plane possible poster practical practicalities presence preserving press probably proceedings processing produces prone propagated propagation proposed pushbroom quality quite radiometric range ranked reason reasons recognition reconstruction reduce reduced reducing references referred refinements remote represent representation result results right rises robust robustly runtime same scene scenes scharstein scholten second seconds section seen segmentation selection semi semiglob sensing september sequence shapes sharp shift shown shows shum siggraph situations sixth small society some space springer stereo stewenius structured structures submitted subpixel symmetric symposium szeliski table taken taxonomy teddy test testing texture than that theory there these they third this through thus time tomasi tool total toward transactions tries truth tsukuba unknown untextured urban using uyttendaele valuable venus very video vienna view viola violation visibility vision visual volume walk wall wang weighted well wells were where which white winder winsconsin with without wrong wrongly yang york zabih zitnick
http://doi.ieeecomputersociety.org/10.1109/CVPR.2006.292	134	Stereo Matching with Color-Weighted Correlation, Hierarchical Belief Propagation and Occlusion Handling Q`ingxioŽng YaŽng Liang Wang Ruigang Yang Henrik SteweŽnius David NisteŽr Center for Visualization and Virtual Environments Department of Computer Science, University of Kentucky	above accurate adaptive after algorithm analysis applications approach approximate april automated based belief birchfield bleyer bolles boykov cartography color colors column comaniciu communications compact compared computer computing cones conference consensus constraints correspondence corresponds cost criterion cuts cvpr cycle data define denoted different disc disp disparity dissimilarity early efficient energy except experiments fast feature felzenszwalb figure final first fischler fitting four from gelautz global graph ground handling hirschmueller http huttenlocher ieee image images incorrect information initial insensitive integral intelligence intermediate international iterations july kanade kang kolmogorov kweon last layered locally machine matching mean measure meer middlebury minimization minimum model mutual neweval nonocc occlusion occlusions okutomi output page paradigm pattern pixel pixels plane planes proceedings processing propagation random rank ratio recognition reference references regions research result results robust rows same sample sampling sawhney schar scharstein search second segmentation semi september sets shift shown shum smoothness society space special stereo supportweight symmetric szeliski table teddy test that theory these third this three tomasi toward transactions truth tsukuba using variable veksler venus visibility vision visual window windows with workshop yoon zabih zheng
http://doi.ieeecomputersociety.org/10.1109/CVPR.2006.76	4	Clustering Appearance for Scene Analysis Sanjeev J. Koppal and Srinivasa G. Narasimhan Robotics Institute, Carnegie Mellon University, Pittsburgh, USA Email: (koppal,srinivas)@ri.cmu.edu	acknowledgements acquiring alexei algorithm algorithms allow analysis appearance approach approximations arbitrary assume authors automatic automation award awards based basri belhumeur beyond bidirectional book books brdf brdfs carnegie change class clustering clusters coleman color colored compute computed computer conclusions conf conference corresponding create curet curless cvpr dana debevec defigueiredo demonstrated derivatives described determining deyoung diffuse dimensional discussions distribution duiker dynamic early eccv effective efficient efros encoded estimated eurographics example exploit extra extrema face field figure four fournier framework from functions general generalization geometrical geometry georghiades ginneken goldman graphics greenberg hanrahan have hawkins hayakawa healey here hertzmann http human hybrid iccv ieee ijcv ikeuchi illumination image images implications including information input intelligence interactive interface intl invariants inverse issues jacobs jain josa kanade kautz klinker koenderink koppal kriegman lafortune lambert lambertian light lighting linear locations machine mallick marschner material materials mccool measurement mellon memo method model modeling motion narasimhan nayar normal normals note novel number object objects observation obtain obtaining only optimization oren outdoor pami paper pattern perspectives photographs photometric photometry physical planes prior proc proceedings processing properties providing radiosity ramamoorthi ramesh real recognition reconstructing recovering recovery references reflectance reflection related rendering report representation research retrieval robotics roughened rusinkiewicz sagar salton sampling sarokin sato scene scenes scia segmenting seitz separable separating sequence shafer shape shashua shree siggraph signal single skin small society source sources spacetime sparrow spatially specular step stereo structure supported surface surfaces surfacesw tabulated tagare tchou technical text texture textured textures thank that theory these this torrance transactions transformation uncalibrated under understanding university unknown using variables varying video vision visual wang westin wheeler with woodham work workshop world zhang zickler
http://doi.ieeecomputersociety.org/10.1109/CVPR.2006.326	41	Using Multiple Segmentations to Discover Objects and their Extent in Image Collections Bryan C. Russell1 Alexei A. Efros2 Josef Sivic3 William T. Freeman1 Andrew Zisserman3	about academy account accuracy acknowledgements affine agency algorithm allocation analysis anna annotation approach area aspects augmented automatic automatically average averaged award bags barnard based basri bayesian berg better bicycles blei borenstein bottom boundaries boundary brandt bray brightness building buildings caltech candidate career carlo cars cast categories categorization category chain class classes classical closely clustering collection color comaniciu combining comparison completely computed computer conclusion conference configurations context corpus correctly correspond covering criminisi csurka cuts cvpr dance data database dataset description descriptors detect detection devel dictionary dirichlet discovered discovering document does driven duygulu each eccv edwards efros evaluated faces feature features fergus figure financial finding finds fitzgibbon fixed forsyth four fowlkes freeman freitas from gatica generates geometric geospatial google gool grant grass griffiths ground have hebert hierarchical high highest hofmann hoiem http iccv ieee image images inspected intelligence intensity intersection intersects into invariant jojic jordan journal karenina keypoints labelled labelme labels large latent learned learning lexicon listing local location locus lowe machine malik markov martin matching measure measurements measures meer memo method methods miller mining minka model modeling models modern monay monte more motorbikes movies msrc mult multiple multiscale murphy names national natural negi news nips nition normalized note number object objects obtained odobez only oped organization other ours outperforms over pages pami parts pattern perceptual perform performance perona pixels plsa power precisions probabilistic proc proceedings project proposed provided quality quelhas ranked ratio recent recog recognition references regions relate report research respectively results retrieval retrieved returned returns roads robots robust russell same scale scene scenes sciences scientific score scores search segment segmentation segmentations segmented segments semantic several sharon should show signs sing single sivic small society spaces springer statistical steyvers such sudderth suggests summarized support systems table takes technical tested text texture than that their theirs then these this tolstoy tool topdown topics torralba training translation trees truth tuytelaars twenty types ullman union universal unlabeled unlabelled unsupervised using verlag video videos viewpoint vision visual vocabulary volume where whereas while white willsky windows winn with within workshop worse zisserman
http://doi.ieeecomputersociety.org/10.1109/CVPR.2006.225	152	Piecewise Image Registration in the Presence of Multiple Large Motions Pravin Bhat1, Ke Colin Zheng1, Noah Snavely1, Aseem Agarwala1, Maneesh Agrawala2, Michael F. Cohen3, Brian Curless1	ability able accurately achieve acknowledgements addition adelson adobe affine agarwal algorithm algorithms alone also ambiguous analysis anandan animation application applications approach approximate areas artificial assessment automated automatically avenues ayer background based baseline because belief belongie bergen better between birchfield bird black bmva body bolles both boykov british camera cannot capture cartography case cases ceramic chum classified clearly color communications compare comprehensible compute computed computer computing conclusion conference consensus consider contain containing continuously correct correctly correspondence correspondences could create created criteria crucial cuts cviu cvpr czech darpa degrees demonstrates dense density depending depth describe desktop detection detectors determining difficult disparities disparity displacement dissimilarity distinctive each eccv editors elephant encoding energy equivalently especially establish estimated estimation euclidean european evaluation even every examples explain explained exposed extracting extremal face fail fails fast feature features fellowship fewer field fields figure figures finally find finds fischler fitting flow found foundation frame frames framework freedom from fundamental future gift given good graduate graph gridded handle handling hanna hard have heavily hierarchical high hingorani homographies homography horn however iccv identify ieee ijcv image images incorrectly independent independently indicated indicates industrial infer information inherently insensitive intelligence interest international intrinsic invariant iterative journal kanade kang keypoints known kolmogorov labels labs large layered layers left levels like likelihood london long lowe lucas machine make many maps marshall matas matching matrices matrix maximally maximum measure metal microsoft mikolajczyk minimization minutes mixture model models more mostly motion motions move moves moving multi multibody multiple national next notice number object objects occluded occlusion only optical optimization other output over pages pair pajdla paper paradigm parameters parametric parts pattern person phantom phenomena piecewise pixel pixels planar planes point portions prague previous proceedings processing produce projective propagation provide pyramid quality random range recognition reconstructed reconstruction recover recovered references regions registration relies representation representing republic required research resolution results right rigid robust rosin rotating runs same sample sampling sastry sawhney scale scene scenes scharstein schmid schunck science search selection september shape shorten show shows shum sift similar single size small smooth soatto society some sparse specular stable static stereo structure such supplementary supported surface symmetric synthesize szeliski table takes taxonomy technique techniques test texture textureless that there these they think this those thus tomasi torr transactions understanding uniform university unlike unoccluded upgrade upper urban used using vectors veksler versa very vice vidal video view views visible vision visualization visualizations visualize visually volume wang washington well went what when where which while wide width wills window with work workshop would zabih zheng
http://doi.ieeecomputersociety.org/10.1109/CVPR.2006.219	115	Particle Video: Long-Range Motion Estimation using Point Trajectories Peter Sand and Seth Teller MIT Computer Science and Artificial Intelligence Laboratory	academic accuracy accurate adaptive algorithm algorithms allocating allowing also alvarez amiaz anandan anisotropic appear appearance approach areas arise artifacts assumption autonomous available away barron based beauchemin belongie better between bilateral black blur both brand brox bruhn camera cases central changes channels chapter cheng cohen coherence color common communication compare comparison complex compression computation computed computer computes computing concatenate conclusion conference consistency consistent constraints construct contour copy correctly correspondence correspondences cost could create current cuts cvpr data decrease deforming delaunay dense deriche described descriptive descriptors detailed detection determined determining didas differs difficulty discontinuities discontinuous displacement distance does dynamic each easily eccv elad enforce enforcing enough enter envision error estimation evaluation existing expect experiment exploiting explore fails feature features feuer fewer fields figure filmmaking filtering first flawed flow focus frame frames fransens from fundamental furthermore future gems geometric geometry good gool gradient graphics ground half handle have high highly hope iccv icip ieee ijcv image improve improves include incorporate incorporating increase incremental interpretation invariant irani isnardi issues jojic justified kang kernel kiryati known lambertian large larger layers like limitation limitations lischinski localization long lower major making malik many mean methods models more morphable most motion multi multiple near normalized objects obtain occluded occlusion occlusions often online only optic optical optimization other others over pages pair papadopoulo papenberg particle particles patches pattern performance permit photo plan point portions position positioning positions possibly press probabilistic problem problems proceedings processing provide provides pruning purposes range ranging rate rather realistic reasons recognition recursive references reflectance region regions rendered replacing representation representations researchers results return reversed rigid robotics robust robustness sanchez santos sawhney scale scene second section segmentation several shift should shown shows silva simplistic simultaneous slow smooth smoothness society some sometimes sophisticated spatial spatiotemporal standard start starting statistical strecha subspace such surveys survival survives symmetrical systems temporal temporally terminated test than that their theoretically theory these they thiesson this thompson thus time tomasi track tracked tracking trajectories triangulation truth unstable used using vector vectors victor video videos vision visual wang warping weickert when will wills with xiao zitnick
http://doi.ieeecomputersociety.org/10.1109/CVPR.2006.306	133	The Panum Proxy algorithm for dense stereo matching over a volume of interest A. Agarwal and A. Blake Microsoft Research Ltd. 7 J J Thomson Ave, Cambridge, CB3 0FB, UK	acknowledge acknowledgements active affi algorithm algorithms analysis approximate artificial available baker based behaviour belhumeur binford binocular biophys blake boundary boykov brown camera circumstances class close clusters computer computing conditions conf conference control coombs corner correspondence correspondences criminisi cross cuts cvpr dense depth discussions disparity dynamic edge editors eigenfaces eklundh energy european evaluation fast fisherfaces fitzgibbon foveated frame from full future gaze graph greater helpful hespanha hingorani ieee image images inproc intelligence intensity interactive interscan intra joint jolly kanade kolmogorov kriegman layer lecture likelihood line linear machine manipulation maximum minimization more multi murray notes number objects obtainable occlusions ohta optimal orientation pages pahlavan panum pattern poggio proc proceedings programming projection proxy pursuit quality quart range ranges real recognition reconstruction references region reichardt reid remains rother scene scenes scharstein science search segmentation shotton smooth society soong specifi springer stereo stringent structure szeliski taxonomy teleconferencing test time torr tracking trans uhlin unconstrained under understanding using veksler verlag video vision visual what with work xation yuille zabih
http://doi.ieeecomputersociety.org/10.1109/CVPR.2006.331	64	Visible Surface Reconstruction from Normals with Discontinuity Consideration Tai-Pang Wu and Chi-Keung Tang Vision and Graphics Group The Hong Kong University of Science and Technology Clear Water Bay, Hong Kong	algorithm algorithms application approach august based bayesian belief believe between bilmes blake boundary brdfs cambridge chellappa cohen computation computer conference correlated curless cviu cvpr dense description detection determining discontinuities discontinuous distortion distributions eccv enforcing estimation expectation field forsyth frankot frey from functionals gaussian geman gentle gibbs given gmip goldman gradient graph graphical hall height hertzmann hidden horn hsieh huang iccv icip icsi ieee ijcv images imperfect inferred integrability january july june karacali koetter kovesi liao maintains markov maximization ment method methods minimizing mirror mixture models modern multiple mumford normal normals noticeable october opteng orientation output pages pami parameter partial pattern petrovic photometric ponce prentice press proceedings produce propagation recognition reconstructing reconstruction references relaxation report representations restoration segmented segments seitz shading shah shape shapelets should snyder society spatially sphere stereo stochastic surface surfaces tang technical tensorial terzopoulos that tutorial useful using varying visible vision visual volume wavelet which with without wong woodham zisserman
http://doi.ieeecomputersociety.org/10.1109/CVPR.2006.37	131	A Planar Light Probe	academic acknowledgements acquiring adaptive addison adviser albedo america anal analysis annual appl application applications approximation arbitrary artoolkit based basis basri belhumeur best bidirectional bmvc brdf brdfs brightness bump california camera capable captured cast catadioptric chellappa choh cipolla color complex comput computer conclusions conf conference constructed cvpr dana debevec december detection device difficult dimensional direction distribution doorn drawing dynamic eccv eigenimages empirical enough environment epstein error estimating estimation eurographics european exposures face february fifth finding foundation fourier fourth framework frequencies frequency from function fundamental further generalization geometry grants graph graphics greenberg greene haar haeberli hallinan hanrahan hara harmonics have high however iccv ieee ikeuchi illuminant illuminants illumination image images imaging intell intelligence interactive inverse investigation jacobs journal june kato kluwer known koenderink koudelka kriegman kweon lambert lambertian light lighting linear little mach machine malik manufactured many mapping maps marschner materials measurements method mitsunaga miyazaki model models most multiple national natural nayar okabe omnidirectional optical order oren other page pages part particularly pattern pbmcv pentland petrov photo photographs photography pitted pixel planar point polarization practical press primitive probe proc proceedings processing projections prone properties publishers publishing push radiance ramamoorthi range realistically recognition recovering references reflectance reflection regions render rendering report requires requiring results right rough samaras sato scene science segal sensitive september shading shadow shadows shape should shown siggraph signal single society sources southern sparrow spatially specific specular spherical springer stanford subspaces such suffice suggests supported surface surfaces synthesis technical techniques technology tedious tern texture that theory thesis this thoroughly torrance trans truly under university unknown useful variable varying verlag view vision volume wang ward washington wavelets weber wesley will with work workshop world would yang yield york yuille zhang zheng
http://doi.ieeecomputersociety.org/10.1109/CVPR.2006.211	62	Nonparametric Priors on the Space of Joint Intensity Distributions for Non-Rigid Multi-Modal Image Registration	about accuracy acknowledgments alignment allows analysis andrew appear approach artificial atsumi based black blake both brox bruhn chapman chefd chen chung clustering collignon column comaniciu comments complete computation computer conclusion conference confirm context convex corporate correspondences cremers criterion cvpr cvrmed data datasets deformable demonstrate density determining directions discussions distance distribution distributions divergence eccv energy enhance entropy estimation existing experimental extension face faugeras feature figure final fisher flow formulation framework frank free from funded funka future gareth gaussian grimson guetter hall hawkes helpful hermosillo high hill horn hornegger hotel huang iccv ieee ijcv illustrates image images include information initial integrate intelligence intensity intrinsic invariant ipmi joachim joint kernel kikinis knowledge kullback larger learned learning leibler level leventon likely london machine maes manuscript marchal matching maximiza maximization mean measure medical meer metamorphs metaxas method methods miccai minimising minimizing modal modality models more motion multi multimodal multimodality multiple mutual nakajima norbash optical order osher outperforms overlap pages pairs pami papenberg paper particular patt pattern powerful press previously prior priors problems proceedings processes proposed provide pure purely qualitative quantitative rasmussen recognition references registered registration regularizers remains research results rigid robust robustness roth sauer schnorr schunck segmentation sets shape shift shown shows siemens silverman soatto society space spatial springer statistical statistically statistics studholme study suetens superior support texture thank that theoretic theoretical theory this tion toward towards training unified using vandermeulen variational viola vision volume warping weickert wells which while williams work zollei
http://doi.ieeecomputersociety.org/10.1109/CVPR.2006.93	150	Coupled Bayesian Framework for Dual Energy Image Registration	alexander american analysis angiography another application april association atkins atsumi ayache bayesian between breast british brown cachier chen chest collignon comparision comparison comput computer computing conditional conf conference considerations correspondence coupled crum csur current cvpr deformations density diagnnostic different digital dual energy estimation fang figure fitzpatrick form framework free from greene hartkens hawkes hayes hill ieee image images imaging information informationbased inria instrumentation intensity international ipmi journal kikinis kinnard kolmogorov leach lncs maes maintz marchal maurer maximization medical meijering method minimization modal multi multimodality multiple mutual nakajima neurological niklason noise nonrigid oliver oxford pages pair pandit patt pattern pdes pluim practice press problems proc proceedings processing radiography radiology rangarajan recog recognition reconstruction references registration regularization regularized report restoration results review rigid rueckert schulze segmentation separated shaw signal similarity slabaugh smoothness society sonoda spie springer subtraction suetens sunku surgeons survey surveys technical techniques theory three trade traditional transactions tyan unal university using vandermeulen viergever viola vision visual volume wang weiss wells whitman zabih zhang zuiderveld
http://doi.ieeecomputersociety.org/10.1109/CVPR.2006.58	25	AnnoSearch: Image Auto-Annotation by Search1	algorithm annotated annotation automatic barnard based blei captioning carneiro centric chang chen cluster color computer conference confidence contentbased correlograms cvpr czerwinski darrell data database detection devices discovery document dumais duplicate duygulu dynamic eccv efficient engineering ensemble faloutsos field fixed forsyth freitas from gcap ghoshal google graph hidden huang ieee image images indexing international ircing jeon jordan journal khudanpur kumar large lavrenko learning lexicon location machine manmatha markov matching microsoft mitra mobile model modeling models multimedia multimodal news nips object pattern photo pictures proceedings quality queries recognition references report research results retrieval scale search searching semantic semantics semi sigir sigmm society technical tollmar tosearch training translation using vasconcelos video view vision vocabularies vocabulary wang with words workshop yang zabih zeng zhang
http://doi.ieeecomputersociety.org/10.1109/CVPR.2006.284	5	Space-TimeVideo Montage Yasuyuki ~atsushita~ Xiaoou ~ a n g t	agarwala agrawala algorirhnz algorithms approximate approximation avrithis ayer boiman boykov coffman cohen colburn computer conference conj conlpuzer content conzpuier conzputer curless cuts cvpr design detecting digital dontcheva doulamis drucker encoding energy estimation extraction fast frames graph gxey ieee ilesign images interactive irani irnugr irregularities johnson kollias layered likelihood lkuns ltzr maximum minimization mixture motion nodelsand oflnt optimal packing pages pattern photomnontage proc proceedings processing recognition references representation robust salesin sawhney scenes society survey sysrenzs updated using veksler video vision volume zabih
http://doi.ieeecomputersociety.org/10.1109/CVPR.2006.221	84	Perceptually-Inspired andEdge-Directed ColorImageSuper-Resolution Yu-WingTai Wai-Shun Tong Chi-KeungTang Visionand Graphics Group The HongKongUniversityof Science and Technology	achieved advantages after algorithms allebach amsderstam analogies analysis approach approaches axiomatic baker based bayesian better blinn blue break carmichael caselles certain chang chuang color comparable complex computational computer conference confidence constraint curless curve curves cvpr december degrees dense derived details digital directed distribution each edge elseviers embedding encoding enforcing enhancement experimentalsection extraction feature field fine framework freeman fundamental geometrically given gonzalez guide hall hallucinate hallucination hertzmann high higher icip ieee image images inferred inproceedingsofieee interpolation into irani jacobs january jvcir kanade learned learning level limit limits local lowlevel magnification makes march matting medioni morel morse motion multi neighbor ness obtained occlusion october offers oliver orchard pages pami pasztor pattern peleg pixel pixels prentice primal priors proceedings process processing recognition reconstruction reconstructionbased references regionness resolution results river saddle salesin same sbert scale schwartzwald science screen segmentation september shows shum siggraph sketch smith smoothness society subject super superresolution szeliski tang tensor tensors term that them these this through todigitalmatting translation transparency typical under upper used using vision volume voting which with within wong woods xiong yeung zheng
http://doi.ieeecomputersociety.org/10.1109/CVPR.2006.102	95	Discriminative Object Class Models of Appearance and Shape by Correlatons	about accuracy additional agarwal almost alone anal analogous appearance approach approaches approximation bags based bayesian been behavior belongie between blake bmvc both bray capri capture captures cartesian categories categorization cipolla class classes classification classifier clearly coding collections color compact computation computer concept conclusions conditions conf conference confusion contexts contour contours conventional corrected corrects correlations correlatons correlograms corresponding criminisi csurka cues cvpr dance database defined demonstrated detection dictionary different discovering discrimination discriminative distributions dorko eccv edelman effect effectively effectiveness efficiency efficient efros enhanced enriched errors european every examples extending extract fast features felzenszwalb fergus fifteen figure finally form fragment freeman from function further furthermore general gray handbook haralick have hierarchical high highlighted hint histogram histograms huang huttenlocher hyperfeatures iccv ieee ijcv illustrated image images improve improves increased increases increasing indexing indicates inform information integral integration intell international intl introduced invariant isolation italy iwvf joint journal julesz june kernel kernels keypoints kumar larger lazebnik lead learned learning left leung lighting local london lower mach made madison malik maps matching maximum means mikolajczyk minka mitra model modeling models multilevel naquet natural nearest neigh neighbour notice noticed number object only optimal outperform over overall pages pami panel panels paper pattern performance perona pictorial plot plots point ponce porikli prague press proc proceedings processing proposed proven puzicha query radius recognition references region regions remaining reports representation respect russell sali saturation scale scene schmid segmentation sensitive shape shotton show shows similar single sivic size society some spaces spatial spot statistical structural structures submitted succinctly such suggests sweet tables tested textons texture that theory there these this three torr trans triggs ullman under universal unsupervised upper used using values varma very vidal viewed viewpoint vision visual vocabulary weber welling where which winn with words workshop zabih zisserman
http://doi.ieeecomputersociety.org/10.1109/CVPR.2006.212	146	Object Pose Detection in Range Scan Data Jim Rodgers, Dragomir Anguelov, Hoi-Cheung Pang, Daphne Koller Computer Science Department Stanford University, CA 94305	acknowledge acknowledgements acquiring action additional agarwal alexander algorithm allow also analysis anguelov appearance approach approximate articulated artificial asme aspects assignments athistos backing based belief berg besl between bhatia bines bioengineering biomechanics black blake body broadly bulow carlsson cases certain closest cloud clutter cluttered coherent color colorado common complex computer concurrent conference configuration configurations consistency consistent consisting constraints context contribution corazza correlated correspondence coughlan could cruz cues cvpr darpa darrell data davis define deformable demonstrate density depth describe descriptors detector detectors determine different directions discriminative distributed eccv efficient elusive empirical enforcing estimating estimation exemplar extension fails fast felzenszwalb ferreira fifteenth finding forsyth framework francisco freeman from frome future gavrila generate generating generation given global graceful graph graphical graphics gratefully gross guibas hand hashing hebert help highly huber human huttenlocher hypotheses hypothesis iccv identify ieee image images important improvement incorporate individual inference informed intelligence intelligent introduce ioffe isard iterative james jitendra johnson journal kanaujia kaufmann koller kolluri kumar lars learn learning level leveraging like likely limbed local loopy loose machine malik mandel matching mckay metaxas method methods metric michael missing mitra model models more morgan mori motion movement munderm murphy nips nonparametric object objects obvious occlusion over pages pairwise pami pampas pang parameter part partitioning parts pattern pauly pearl people performs pictorial place placement point pose poses presence prior probabilistic proc proceedings program propagation provides prune ramamoorthi range ravi real reasonable reasoning recognition recognizing recovering references regional registration regression relevance richer roth rusinkiewicz santa scan scans scene scenes sclaroff segmentation sensitive sets several shakhnarovich shape shapes sigal significant silhouettes sminchisescu society space spacetime spin srinivasan stanford stefano stereo structure structures study successfully such sudderth sullivan summer supported surface surprisingly survey symposium systematic systems szymon texture that then there these this those thrun torr toyama tracking transactions transfer treatment triangulation triggs uncertainty under understanding unifying unlikely unsupervised using utilizes vail valued variability vector viola vision visual weiss well when which willsky with work would xiaofeng zhiguo zisserman
http://doi.ieeecomputersociety.org/10.1109/CVPR.2006.251	138	Region-Tree Based Stereo Using Dynamic Programming Optimization	acknowledgements alberta algorithm algorithms also analysis anonymous approach april august authors based beijing better birchfield blake bleyer build camera canada chen china choi christoudias city comaniciu comments comparison computation computer computing conf conference consistency constrained constraints constructing control cooperation copenhagen cormen correspondence criminisi cuts cvpr deng denmark dense department diego disparity dissimilarity dynamic edition ernst european evaluation explore fast feature figure financial frame framework france from furthermore gaze gelautz generalized georgescu global gong graph ground handling harpreet hong http icore ieee image insensitive intelligence international introduction issue journal july june kambhamettu kang kolmogorov kumar layered lerserson level like machine manipulation matching mean measure meer method middlebury model motion multi multiple nice nserc occlusion october olga optimal overveld pages pairs pass patch pattern photogrammetry pixel points ponald press proc proceedings programming progressive purpose quan quebec rakesh recognition reconstruction references region reliability remote results reviewers rivest robust sampling sawhney scene scharstein science segment segmentation sensing shift shotton shum society space spie stein stereo structure supports surface symmetric synergism szeliski tang taxonomy teleconferencing thank that their tomasi topic torr toward trans tree trees university using valuable vancouver various veksler visibility vision volume washington wilinski with worthy would yang zabih zhang
http://doi.ieeecomputersociety.org/10.1109/CVPR.2006.54	97	An Integrated Model of Top-down and Bottom-up Attention for Optimizing Detection Speed Vidhya Navalpakkam Laurent Itti	america anticipation appearance applied artificial attention ballard baluja based bauer between biol bottom bouchard boundary braithwaite bulletin butcher cars castelhano categorization category change changing cognit cognitive color colour combination comput computer conditions conference context contribution contributions control cortex covert cowan critical culhane curr cvpr davis detecting detection determined differences dimension dimensions discriminability distractor distractors down driving duncan edgington effect effects efficient elec evidence exhaustive face faces fast feature features fergus foreknowledge from gain gelade guidance guided hayhoe henderson hierarchical hodsoll horowitz human humphreys hyle icip iconic ieee imaging increases influences inhibition integration intelligence itti jiang jolicoeur jones journal kanade kenner king koch learning linear linearly macaque maljkovic martinez mechanism method mind model modeling moghaddam motion movements nagy nakayama nature navalpakkam networkbased neural nips nuflo object objects oliva optical optimal overt pages pami papageorgiou parallel part pashler pattern patterns pedestrians pentland percept perception perform perona poggio population preview primate priming probabilistic proc proceedings processing psychol psychological psychology psychophys psyonomic real recognition references relative representation research responses review revised robust role rowley rutishauser saliency sanchez schneiderman search selection selective selectivity separability separable september setting shifts similarity singletons sinha size snow society sparse speed statistical stimulus strategies strategy system systems target targets task template templates that theory time torralba tracking trainable trans treisman treue triggs tronic trujillo tsotsos tuning underwater useful using vasan vickery video viola vision visual volume walther with within wolfe your zelinsky zisserman zmura
http://doi.ieeecomputersociety.org/10.1109/CVPR.2006.51	23	Aligning ASL for statistical translation using a discriminative word model Ali Farhadi, David Forsyth Computer Science Department University of Illinois at Urbana-Champaign	above advances afgr alignment allow american analysis appear application applications approach assan base based bauer below between bilingual bitext block boundaries bowden brady breaking brown budget building bungeroth cambridge casterline center channel chart cocke combined compared complex computational computer conf conference continuous corpora coupling critique croneberg curin cvpr data dealt derived desk dictionary dinosaur diplomat discriminative discussion distinctive down dynamic dynamical eccv empirical encoding english enough errors explicitly exploiting fact feature features figure film final finite forests foundations framework frederking from future gait gallaudet general generation gesture good grammar grandpa grishman grobel handshapes hannan have here hidden hienz hmms huenerfauth icsmc ieee ijcv illustrate image impact information interactive interpretation into invariant involve isolated jahr jelinek kadir keypoints knight kosaka lafferty language large liddell limited line linguistic linguistics lowe lrec machine macklovitch manmatha manning markov matching mean meaning melamed mercer metaxas method methods meyers model models more motion movements multiple natural nice notice novel november onaizan order pages pami parallel pattern penn pentland phonemes phrasal phrase phrases pietra possible posterior press proc proceedings processing produce productive project purdy quite rath real recognition references relatively relevant replacing report representation results roossin rule scalability scale schut sentences sequences shaded shared shown sick sign signs simple size small smith society sophisticated speech spotted spotting starner state statistical stokoe straightforward structure suggests support surprisingly survey systems taking technical technology terms texts that their then this tight time together tools toward transfer translatable translation univ university using valley vector very video vision visual vocabulary vogler warping washington wearable weaver whose will windridge with without word words wordspotter work workshop wrong yangarber yarowsky your zisserman
http://doi.ieeecomputersociety.org/10.1109/CVPR.2006.12	117	3D Building Detection and Modeling from Aerial LIDAR Data	acknowledgement ahlberg airborne alharthy applying approach approaches architecture archives automatic baillard ball based bethel brazilian brenner building buildings city computer conf conference constructing contract cvpr data debevec density elaksher elmqvist environments extraction faces feature filho filtering fitzgibbon from fruh geoinformatics geometry graphics ground haala height heuristic hierarchical high hybrid iccsa ieee image interactive international isprsgis july large lidar line lopes malik matching medeiros merging modeling models multiple neumann nevatia number page pages pandrs pattern persson photogrammetry photographs pivoting planar plane pmtrasys proceedings processing recognition reconstructing reconstruction references remote rendering representation research resolution restricted scale schmid sensing siggraph site society soderman supporting symposium taylor thank this trees urban using velhoa very views virtual vision visualization vosselman xvii zakhor zisserman
http://doi.ieeecomputersociety.org/10.1109/CVPR.2006.73	60	BoostMotion: Boosting a discriminative similarity function for motion estimation	accuracy achieved acknowledgement active additional additive algorithm alison allen also appearance application approach artificial avidan based because black blake block boosted boosting bootstrapping boukerroui brady brought brox bruhn cascade choudhury cipolla classification cohen collins comaniciu combining compared computation computer conf conference constant contour contours conventional cvgip cvpr darpa data decision detecting detection detector determining dinstein discriminative distances done drifting eccv echocardiographic efficient enhanced ensemble error estimates estimation experimented face fast feature features finally flow framework freeman freund friedman function functions fusion future general generalization good guided gupta hastie have high horn iccv ieee ignores image images imaging include incorporating information intelligence intl investigate ipmi iterative jacob january jones journal kanade krishnan krucinski large leaning learned learning lepetit likelihood line location locationsensitive logistic lower lucas manner margin matching maximum mean measuring medical meer mikic mikolajczyk modeling more motion multiclass murphy negative negatives noble noisy nonparametric object objects online optical oren outperformed pages pami papageorgiou papenberg parameter part pattern patterns pedestrians perspective piecewise pilet poggio point pose positives power probabilistic problem proc procedures proceedings proposed providing ramesh rapid real recognition reduced references registration regression regular requirement responses rigid robust roth savings schapire schemes schmid schunk sciences segmentation selection sensitive sequence sequences shao shape sharing shift siemens similarity simple singh smoothing snow society solutions sparse spatial sriram statist statistical statistics steerable stereo storage structure support supported system technique temporal test thank their theoretic theory thomas tibshirani tied time tomasi torralba track tracking train training trans ultrasound understanding unified update using vector velocity video view viola vision volume warping when which wiecker will williams with work works workshop zhou
http://doi.ieeecomputersociety.org/10.1109/CVPR.2006.204	127	MultiviewGeometryfor Texture Mapping 2DImagesOnto 3DRange Data	about accuracy acquisition aerial affine alignment allen anal analysis applications approach approaches approachprovidesa astrom australia automated automatedfeature automatic automatically automaticregistrationof based baseline beardsley benefits besl between bolles borne buddha calibration cambridge camera cameras canada cartography cass ccny chen christy city clouds cnrc colored comput computation computer concalves conference consensus constant constructing continuous cornelis correspondence correspondences cvgip cviu cvpr data davis dementhon determination diego digital distinctive dornaika dpvt duchamp dwith edition efficiency environment environments error estimationusingcoplanarfeaturepoints euclidean factorization faugeras feature features figure fischler fitting flexible from fruh full gaussian generality geometric geometry geosystems gool graphics great ground group hall handheld hartley hausler heyden high horaud html http human huttenlocher iccv icpr ieee ijcv ikeuchi image imagery images imagetomodel imaging increased information intell intelligence interaction interior international intl intrinsic invariant invitation ismar iterative jacobs january japan journal july june jurie keypoints koch kosecka lamiroy large laser leica leordeanu line linear link localization locations lowe luong mach machine mapped march matching mckay merged merging method methods metric minimal model modeling models motion multiple multiview murray ndieeeworkshop nister notice november oberkampf object objects onto onvideo ottawa pages pami papadopoulos paradigm parameters paraperspective patt pattern perspective photorealistic point pollefeys polynomial pose positions press previous problem proc proceedings processing production project projective pulli quan rached random range real reality recognition recognizing reconstructing reconstruction recovered recovery references regions registration rendering resolution respect result ritter robustness sample sastry scale scanners scene scenes schaffalitzky second self semi separated sequeira sequential seven shadow shapes shapiro shepard shibasaki shown simultaneous single soatto society solid solution space springerverlag stamos stamosandm stamosandp statistical stereo stratifed structure stuetzle sujet technique technology texture textured time tokyo tops trans transactions triggs troccoli tuytelaars ullman underst university updating urban using vehicle verbiest vergauwen video view viewpoint views vision visual visualization volume weak wells wide widely with world zakhor zhang zhao zisserman
http://doi.ieeecomputersociety.org/10.1109/CVPR.2006.199	141	Multi-ViewStereoRevisited MichaelGoesele BrianCurless StevenM.Seitz UniversityofWashington	about accuracy acknowledgements adobe alexander algorithm algorithms andanendowment animation application applicationforwhichthemodelisgeneratedand areas around available based baseline behaved best bestmethod better between building byrobshortandemerdooley camera carved channel cipolla circular cluster color column comparison completeness complex compute computed computer conference consistency constraints constructing cornelis correlation correspond cross curless cuts cvpr data dataset datasets dense diebel difference differenceto dimensional dino dinofull dinoring dinosparsering directly dynamic eccv ecole ecommunications estimated evaluated evaluation even evolution exact faugeras fellowship feodor figure firstcomputetheaveragecolorvaluevi flow foreachvectorwith formulation foundation from full furukawa fusion geometry given gool grail grant granted graph ground hand heights held hernandez highest hingorani hours http hulls humboldt iccv ieee ieure image imagebased images information international interval italics jean journal kanade keriven koch kolmogorov labs left level levoy like likelihood lists little lncs local lynen maximum means method methods metric microsoft minutes model modeling modelling models modern most motion multi multiple multiplication mview narayanan nationale normalization normalized nskulla object objects okutomi orofauser other package pages paris part pattern pentium performing philosophy photo photograph photography pixel plausible pollefeys ponce pons position preference principles prior problem proc proceedings processing processor product progressive providing quan question rander range real recognition reconstruct reconstruction references regarding registering research results returns right ring runtime scene scenes scharstein schmitt seitz sequences siggraph silhouette sillion single sinha smile smooth society software sparsering stereo super supported surface szeliski table temple templefull templering templesparsering thank therefore thesis third this time times timings tops torr tpami trans trast truth uiuc ultimately uncalibrated under understanding university used using value values variational vectors verbiest vergauwen version view virtual vision visual vogiatzis volumetric vrip vrippack washington wecanthencomputethenccinastandardway well were where whether whose window with work workshop worlds would yasutaka yfurukaw zabih zeng
http://doi.ieeecomputersociety.org/10.1109/CVPR.2006.97	142	Depth from Familiar Objects: A Hierarchical Model for 3D Scenes Erik B. Sudderth, Antonio Torralba, William T. Freeman, and Alan S. Willsky Department of Electrical Engineering and Computer Science Massachusetts Institute of Technology	accurately acknowledgments affine agency also amer analysis appearance applied approximately arda assoc background bags based baseline bayesian beal believe berkeley blei bmvc bookshelf bottom carlin categories categorization center chain chapman chen chum chung class clockwise coarse comp computer conference context contextual corresponding csurka cues cvpr data dataset density depth describing desk detect detection detectors developed dirichlet discovering discovery discussion distinctive each eccv efros entropy escobar estimates estimation exciting explore extremal familiar feature features fergus figure flexible foundation framework freeman frey from funding gelman generative geometric geometry geospatial grant graph green group hall have hebert hierarchical histogram hoiem iccv ieee ijcv image images improve improves incorporation infer inference inferred instances integrated intelligence interest into invariant jojic jordan june keypoints layers lazebnik learning left less location lowe markov matas maximally maximum meters methods mikolajczyk mixture mixtures model models monitors monocular more multiple national neal near negi nips object objects office oliva overlay pages pajdla pami parsing part parts pattern performance perona perspective point ponce presence press proceedings process processes program projections provide provided recognition references regions report result right robust rubin russell sampling saxena scale scales scene scenes schmid science segmentation segments show single sivic society solely sprites stable stat statistical statistics stereo stern structure structured sudderth systems technical test texture than that their these thus torralba trans transformed unifying unsupervised urban using vace video viewpoints vision visual volume west wide will willsky with workshop yuille zisserman
http://doi.ieeecomputersociety.org/10.1109/CVPR.2006.317	113	Tunable Kernels for Tracking Vasu Parameswaran, Visvanathan Ramesh, Imad Zoghlami Real-Time Vision and Modeling Department Siemens Corporate Research Princeton, NJ 08540	algorithm aligned allowing analysis anisotropic appearance approach areas artificial bandwidth bandwidths based bebis birchfield blob blocks bottom chose classic cohen collins color colorhistogram comaniciu computer conf conference content correlogram cvpr data davis decide dewan driven duraiswami each elgammal estimate european exploring feature features figure finally fleet four frames function fusion future given grossberg hadjidemetriou hager histograms huang ieee image improved intelligence intelligent international intl issue jepson joint july kernel krose like line machine maraghi mean meer method miller model models motion multiple multiresolution nayar number object objects online open optimal pattern positions probabilistic proc proceedings proposed ramesh rangarajan real recognition references region representation retrieval robust rotation same scale segmentation selection shape shift society space spaces spatial spatiograms srihari state stewart systems target their these thiesson this through tools toward tracker tracking trans translation transportation using vehicle versus vertically video vision visual wang will with zhang zhao zivkovic
http://doi.ieeecomputersociety.org/10.1109/CVPR.2006.308	118	Three-dimensional Volume Reconstruction Based on Trajectory Fusion from Confocal Laser Scanning Microscope Images	academic across addressed adjacent alignment analysis appropriate attitude bajcsy base based benson bielke biology biomedical blackwell border bryan caspi cell cells chan characterizing choice chung clsm clutter cohen complexity computer computing concern conclusion conference confocal constants continuity continuous cornell cost could curve cvpr decision decisions defined degree depth designing diagnostic diego digital dimensional discontinuity discuss dorst dotted driven during efficient eliminate error estimating estimation extracellular feature finally first fitting florida flourescent fluorescence fluorescent folberg from function fusion global gotto guided heterogeneity hybrid ieee image imagery images imaging important individual information inside intelligence interest international introduction investigation ipmi irani journal jungke kang knowledge known labeled laser lines lscm machine main matched matching matrix medical medioni meindl melanoma method methods metrics microscope microscopy minimization minimized model morphological most motion multiple multiview optimal order orthogonal over parameters pattern performing photobleaching plant plotted point polynomial preserve preserves press priori priorities problem proc proceedings processing procrustes projections propagation proteins rate recognition recommended reconstructed reconstruction references registration research residual salient scanning second seelen select selection sequence should smith smoothness society solid spatial spie step stevens structures studies subvolumes supporting symposium synergy system systems tensor terms then these thorough three threedimensional tissue tomography tosequence towards tracking trajectories trajectory transactions transformation trochim university user using uveal vascular video vision volume volumes voting wbir weighting which within workshop
http://doi.ieeecomputersociety.org/10.1109/CVPR.2006.24	80	A Framework for Feature Selection for Background Subtraction	able adaption adaptive additional addresses adverse after algorithm algorithms amsterdam analysis analytically application applied approach artificial august azarbayejani background based bayesian been better body boosting boult brandt brumitt camera cappellini changes chapter characterization choice choose classification classifier clear coetzee collins color comaniciu combination combining compensated computer conference confidence constraints contribution copenhagen cvpr darrell davis decision denmark density different dimensional dual dublin dynamic eccv effect elgammal elsevier error estimation european examples experimental experimentally facilitates feature features feedback filter first florida follow foreground framework france freund friedman from future gaussian generalization generalized gradient gradients greece greiffenhagen grimson harville harwood head hierarchical high higher hilton however human iccv ieee image improved inevitable information input intelligence interactions international interpretable introduces ireland island javed journal july june justifiable kalman karmann kerkyra kernel koller krumm learned learning level like line machine maintenance malik memory method methods meyers mittal mixture model modeling models monnet more motion moving multiple negative netherlands nice niemann object objects occlusion oliver others outdoor page pages pami paragios parametric pattern pentland perform performance perpixel pfinder pless practice predictions previous principles probabilistic problem proceedings processing proposed ramesh rated real reasoning recognition references robust rosario russell satisfactorily scene scenes schapire sciences sclaroff segmentation segmenting sense sept sequences setup shafique shah shown signal singer society some spatial spatio statistical stauffer stockholm structural subtract subtraction surveillance sweden system temporal terms textured than that theoretic there thirteenth time toyama tracking tried uncertainty unless using varying video vision volume wallflower weber which will with work wren zhong
http://doi.ieeecomputersociety.org/10.1109/CVPR.2006.124	10	Fusion of Summation Invariants in 3D Human Face Recognition Wei-Yang Lin Kin-Chung Wong Nigel Boston Yu Hen Hu University of Wisconsin-Madison Department of Electrical and Computer Engineering 1415 Engineering Drive, Madison, WI 53706	about acoustics acta affine algorithm algorithms alone along also analysis applicandae application applications approach approaches area areas audio austin austria authentication authentification automatic based baseline beginning biometric biometrics blackburn bone boston bowyer bronstein cartoux cases challenge chang cheeks chellappa chiaki cipolla coframes combination compute computer computing conf conference consuming correlation current currently curvature curved curves cvpr data depth designed development devices differential dimensional direction directions efficient efforts eigensurface either erlebacher errors evaluation experiments expression extraction eyebrows eyes face faces failure features fels feret figure flynn foundations frgc from future general geometry gesture gestures good gool gordon grand grother hann hausdorff have heseltine hesher hickman hoffman human ieee ikeda image images imaging important information integral intelligence interpretation intl invariant invariants issues journal june kakadiaris kempenaers kimmel koch konstantinidis lapreste like limit literature little machine mapped marques matching mathematicae medioni methodology micheals milios minolta minutes missing modeling moon moons more most moving multimodal murtuza needed normalization normalized note novel object obtain obtained olver oosterlinck overview page pages paper part passalis pattern pauwels pears person phillips physiological pictures practical principal proc proceedings processing produce profile projective provided quality range rate rauss recognition recovering reduced references regions regularization reliably reported research rest results richetin rizvi rosenfeld rotations russ same sato scenes scruggs semi sensing seventh shape should signal society some special speech spherical srivastava step such summation surface survey surveys symposium tabassi take takes tanaka technique technology tend test that theoharis theoretical theory these third this three thus time toderici trans translations unaffected unnormalized using values vendor verification very video vienna vision vivid volume wang waupotitsch where which with worek works workshop would wrong zhao
http://doi.ieeecomputersociety.org/10.1109/CVPR.2006.266	149	Scale-driven Iterative Optimization for Brain Extraction and Registration	absolute alliney also always apparent around available background because benefit besides best between both brain brainweb cannot cases clear color computer conference cvpr data digital does doing each empirical estimation evaluation existing extraction filters finding foreground from future general gold good gray however ieee image images improve improving include information initial initialization intensities knowledge largest method modalities model more mostly norm obtained only other pattern plan present problem proceedings processing proposed recognition references regarded regions registration regularizers reported results robust scale second section segmentation show signal significantly similar since single situations society solution standard strong such systematic that them there thesholding this those timation tissue trans utilizing very vision weighted well where whole work works would
http://doi.ieeecomputersociety.org/10.1109/CVPR.2006.206	61	A New Method of Probability Density Estimation with Application to Mutual Information Based Image Registration Ajit Rajwade, Arunava Banerjee and Anand Rangarajan, Department of CISE, University of Florida, USA.	accurate affine algorithm aligned alignment already although analytic another application approach arbitrary asilomar based between brady brain british ccre challenges chen collignon collins comput computational computer computers conference consequence construction continuous cost costa cover cumulative cvpr davis deal density derivatives described design difference differentiability differentiable digital distributions does duraiswami efficient elements entropic entropy essential estimating estimation estimator exploit exponential fast finding finds flat form formulation front further furthermore future gauss gorman gradient gradients graphs grimson group gumerov have hero iccv icip ieee image images imaging immediate implementation improved information intensities intensity interest interesting interscience involve issues joint kadir kernel latter learning leventon louis machine maes main manifold marchal maximization measure medical methods metric miccai michel minimum modal multi multimodality multiple mutual note number overall pages paper pattern phantom pose positioning present presents proceedings rangarajan realistic recognition references regions registration relative required research residual rigid sacrificing sampling search section signals society spanning statistics suetens switching systems technique that theoretical theory these this thomas though trans transactions transform tree unfortunate using vandermeulen vemuri viola vision volume wang wells which wiley wise with without work would yang york zhang
http://doi.ieeecomputersociety.org/10.1109/CVPR.2006.75	50	Classifying Human Dynamics Without Contact Forces	abruptly achieve acknowledgements acoustic actions adaptive afosr agarwal algebraic allerton also analysis angles appears application applied appreciate approach approximate approximated approximation arma articulated assume automatic automatica autoregressive available based benabdelkader between beyond bilinear biological bissacco blake bobick bodenheimer bowyer boyd bregler brien brostow camps canada capture challenge change changing character chellappa chiuso class classification classify close closed coch comp complex computable computer concentrate conference contact content context contr contrary cuntoor cutler cvpr data databases dataset davis december decomposition dependent detecting detection devise difficulty discriminated discriminative discussion displays distance distances distributions does doing dowson dptv drawing dvir dynamic dynamics eccv efficiently elgammal enough estimates estimation euclidean eurasip event eventually experiments exponential extracted face february feedback figure filtering form frechet from gait gaits gesture good graphics greenspan grimson grother gustafsson hall hard have hidden hodgins human humanid humans hybrid ideally identification ieee image images impact improvements inference inferrable inspired interface intl into introduce irani isard issue jain johansson joint journal juang june kale kluwer korea krueger landau learned learning like likelihood limp limping linear literature little ljung localizing maccormick magnetic malik manner manor maps markov martin matching math maximum mazzaro mean means metric mixture model modeling models moment montreal moor motion movement multivariate murray networks nonetheless normal north novel optimistically order over pages pami parameter parameters pattern pavlovic people perception perform performance performances periodic perona phillips posterior prague prefer prentice presented primitives principled problem proc proceedings processes processing propose psychophysics quasi rabiner rajagopalan recognition recognize recognizing references rehg relies resamplig research resort restricted results rittscher roychowdhury rubner running sakkary sarkar segmentation self seoul sept sets shah shape showed shown shows signal signals significant similarity simple simpler since soatto society solution somewhat sons space speech sponsored stationary statistics style subspace successfully sundaresan support surprising symp system systems sznaier tasks technique techniques templates temporal that their theory these this tracking trans transient treadmill triggs tugnait twists under understanding unstable used user using validation varieties vecchio vega very vidal video videre vision visual volume walk walking wasserstein wealth which wiley with would yields zames zelnik
http://doi.ieeecomputersociety.org/10.1109/CVPR.2006.290	74	Specular Flow and the Recovery of Surface Structure Stefan Roth Michael J. Black Department of Computer Science, Brown University, Providence, RI, USA	accuracy acknowledgments adapting algorithm analysis anandan appearance application applications arvo assisting available avidan ayer barron based beauchemin bergen besl binford black blake bouguet bouguetj brain brelstaff brox bruhn bulthof burt calib calibration caltech camera capture chen common compact component computational computer computing conference containing cornelius criminisi cviu cvpr data depth determining diffuse discussions does dominant domini dpvt dyer dynamic eccv editor estimating estimation exchange experiments extended extraction fields fitting fleet flow framework from fulvio gazette geometry giblin graphical graphics haytham healy helpful heyden high hingorani hinton http hybrid iccv ieee ijcai ijcv ikeuchi image images incremental industrial information intelligence interd international irani jain jordan journal justifies kanade kang kluwer know lambertian layer layers learning local machine mase math mathematical matlab measure method mirror models monthly moscovich motion motions moving multiframe multiple nature nayar neal neumann object observer optical oren orientations other pami papenberg parallax parametric part path pattern peleg perception performance perona perturbation photometric physics piecewise presence principle proceedings processing qualitative recognition reconstruction recovery references reflectance reflection reflections remarkable representation robust roth sarel savarese sawhney segmentation separating separation sequences shading shape shizawa shum sigart siggraph smith smooth soatto society solem sources sparse specular specularities specularity spherical stereo superposition supported surface surfaces swaminathan szeliski techniques thank that theory this threeframe through tomer tong toolbox transactions transparency transparent using variableorder variants variational videos view vision visual waldon warping weickert with work workshop yezzi zisserman
http://doi.ieeecomputersociety.org/10.1109/CVPR.2006.146	35	Image-based multiclass boosting and echocardiographic view classification S. Kevin Zhou, J.H. Park, B. Georgescu, C. Simopoulos, J. Otsuki, and D. Comaniciu	across additive affine always amit application approach arrangement artificial automatic bakiri based bayesian between boosted boosting bottou cascade categories chang classification classifier clustering codes computer conference confidence correcting cvpr deal decision descriptors detect detection dietterich discriminative duncan ebadollahi efficient elements error evaluate evaluating event face feature features fergus floatboosting framework freeman freund friedman from future geman general generalization generic hastie hopkins huang iccv ieee improved increase inechocardiogram intelligence invariance invariant jaffe johns jones journal krempp lazebnik leaning learning lecun lighting local logistic lowe machine method methods models multiclass murphy neighborhood number object objects online oren output outputs pages pami papageorgiou parts pattern perona plan poggio ponce pose predictions probabilistic problem procedures proceedings proposed rapid rare rated recognition references regression relation report representation reusable scale schapire schmid sciences section sequential sharing shot similarity simple singer society solving spatial springer statist statistical statistics structure such system tagare technical texture theoretic tibshirani ties tomograhic torralba training tree univ unsupervised using videos view views viola vision with zhang
http://doi.ieeecomputersociety.org/10.1109/CVPR.2006.250	98	Region-based Image Annotation using Asymmetrical Support Vector Machine-based Multiple-Instance Learning	algorithm algorithms also analysis andrews annotated annotation annual appearance approach artificial asvm automatic axis barnad barnard based bayesian benchmark bernoulli blei both canada carneiro categorization chang chen competitively complement components computer concepts conclusion conference content cross cuts cvpr data dietterich diverse dividing dominant dong duygulu eccv especially evaluated even experiment experiments failure feng first fixed formulating forsyth fotouhi framework freitas fretias genova greatly hofmann icip icml ieee image images improve instance intellegent intelligence international interpreted intetational italy jeon jordan journal keywords lathrop lavrenko lazy leading learning level lexicon library libsvm lozano machine machines malik management manmatha maron matching media methods misvm modeling models mori multi multimedia multiple multipleinstance musk natural nips normalized novel object optimized outperforms pages paper parallel particular pattern perez performance pictures porc problem problems proc proceedings proposed puerto quantizing reasoning recognition rectangles references region regions relevance research result results retrieval rico runs scenes segmentation semantic semantics sets should show shows sigir singapore society solving storage superiority supervised support svmbased takahashi that they this those though through toronto transformation translation tsochantaridis using vasconcelos vector very video vision vocabulary volume wang with word words workshop yang zucker
http://doi.ieeecomputersociety.org/10.1109/CVPR.2006.17	147	3D Surface Matching and Recognition Using Conformal Geometry Sen Wang, Yang Wang, Miao Jin, Xianfeng Gu, Dimitris Samaras State University of New York at Stony Brook Computer Science Department Stony Brook, NY 11794, USA {swang, yangwang, mjin, gu, samaras}@cs.sunysb.edu	abidi advances aggarwal alliez alon amer analysis analyze angenent angle application applications approximate arbitrary athitsos atlas automatic available average based become belongie best boostmap both bowers brain bulow cambridge campbell cerebellar chan chazelle chen choice chua classification cluttered compatible computer computers computing conclusion conference conformal contexts continue coordinate correa cross curvature curve cvpr data database databases deformations densely derose desbrun descriptors design different differential disk distinctive distributions dobkin donamukkala duchamp eccv eells efficient elementary engine engineering eurographics evaluated exploit face faceted family features figure fingerprint first flat flattening floater flynn form framework free from frome fully funkhouser further furthermore future generation genus geometric geometry global gool graphics greiner gupta haker halderman halle harmonic harvard have hebert high hoppe hormann huang huber hurdal iccv ieee ijcv image images imaging including information institute international intrinsic invariant jacobs jarvis johnson june kapuria kazhdan keypoints kikinis koch kollios kolluri kraevoy larger least lectures levy lounsbery lowe maillot malik malo manifolds mapping mappings maps matching math medical meila meshes meshing method meyer mips mitiche modelling models mori motion multiresolution mumford neill neuroimage noise novel object objects occlusion only osada pages pami paper paradigm parameterization parameterizations parametrization parts pattern performance petitjean plan point presented press problem proceedings processing proesmans properties proposed range rankings recognition recognizing references regional registration remeshing representation representations resolution resolutions result results riemannian rigid robotics robust rotation rottenberg ruiz rusinkiewicz saint samaras sampled sampson sapiro scale scenes schoen sclaroff search second shape shapes shapiro sharon sheffer shown siggraph similarity simplified society spherical spin springer squares stephenson stuetzle sturler subjects such sumners surface surfaces survey symposium systematically systems table tannenbaum technique techniques texture that their thesis this thompson topology tracking transaction transactions tutorial under understanding univ using validate variation vemuri vision visualization volume wang were which will with work wyngaerd yaun zero zhang
http://doi.ieeecomputersociety.org/10.1109/CVPR.2006.273	72	Separation of Highlight Reflections on Textured Surfaces	accurate acknowledgement after ambiguities ambiguity america analysis appearance applications based basis beyond boult breneman brings captured chromaticity closeup color complete completely components comprehensive computer conf conference consequently consideration constancy constrained correlation could cross cvpr decomposition depth diffuse diffusespecular dimensional discrimination distributions early efros entropy european evident fang fields figure filters final finlayson frame framework freeman from functions future general grant graphics ground handbook highlight highlights hkust hong hordley however hubel ieee ikeuchi illumination image images improvements information inpainting instances insufficient intell intensity international inverse investigate jain journal kanade kang klinker kong leung liang light linear machine malik matches materials maximum measurement mechanisms method model modeling models more mumford nayar nishino observed obtained optical original pages patch pattern perona plan polarization potentially preattentive proceedings processing proposed provide quan quilting random realtime recog recognition recognizing recovery references reflectance reflection region removal representation representing research resolution resolve results sampling sato scales schulte separate separating separation sequences shafer shum siggraph simple single smoothly society some space sparse spatial spatially specular statistical supported surface surfaces synthesis technique temporal textons texture textured textures theory three through throughout tiple tong towards traditional trans transfer truth tuceryan unified unifying using variation varying vision visual which windows with work
http://doi.ieeecomputersociety.org/10.1109/CVPR.2006.228	33	Probabilistic 3D Polyp Detection in CT Images: The Role of Sample Alignment Zhuowen Tu1, Xiang Sean Zhou3, Adrian Barbu2, Luca Bogoni3, Dorin Comaniciu2	acar achieves acknowledgements adaboost additive aided algorithm algorithms alignment also amer analysis anatomical appearance appendix application assisted automatically based beaulieu before bogoni boosting bound cancer cancers candidate cascade characteristics chen classification classifier classifiers cleansing cleynenbreugel clin clustering colon colonic colonography colonoscopy comaniciu comp complex computer computeraided computing conference confidence consuming curvature cvpr databaseguided decision defined dept detecting detection diagnosis differential dimensional directly discriminative distinctive does down endosc error explicit face fast features floatboosting freund friedman from function gastrointest gastrointestinal generalization generation georgescu gktrk gourdon gupta hastie have heterogeneous iccv ieee image images imaging improved incorporated inside intervention into invariant isointensity jeffrey jerebko jones keypoints kiss know krishnan laufer learned learning left line logistic lowe lung lymph machine marchal materials medical method models multi napel natural needs nodes nodules normalization nppi number paik pami paper partition pattern patterns philadelphia physical polyp polyps predictions probabilistic proc procedure proceedings processing pulmonary radiology rated rather reading recognition references regression remove report require requires residual result robert sample saunders scale scenes schapire scheme segmentation selected sept shapes shown side siemens singer society spin stage stanford statistical statistics step structures such suetens suitable surface surfaces symmetric task technical text textbook than thank then theoretic therefore these thirion this thoeni though three tibshirani time tomasi trans tree undergoing understanding univ using verifying view vining viola virtual vision weak with without yaov yoshida yuille zhang zhou
http://doi.ieeecomputersociety.org/10.1109/CVPR.2006.231	56	Pursuing Informative Projection on Grassmann Manifold	accelerated achieved acknowledgement administrative algorithm algorithms anal appendix appl arias artificial assessments bayesian because belhumeur benavente boosting characteristics chinese class classification classifying combine communications computer conducted conference constraints convincingly council cvpr data database described determined deterministically discriminant diversity duda edelman edition eigenfaces enhanced evaluation exploit extended extensive extraction face fano feature features feret fisher fisherfaces framework from fully geometric geometry grants hart have hespanha highdimensional hong iccv ieee improvement information informative intelligence interscience journal kong kriegman kullback learning leibler linear luettin machine maitre manifold mann marginal martinez matrix maximization maximum methodology models moghaddam moon mutual naquet nips numerical object optimization orthogonality pami paper parametric pattern pentland perceptual philips practical problem proc procedure proceedings projection proof proposition protocol proved purdue rauss recognition references region report research ryzvi selection show shum siam significant significantly simth society solution space special specific statistical stork subspace supported tang tasks technical theory this torkkola trans transmission turk ullman unified uniquely univ university using vasconcelos vidal vision wang wechsler wiley with work york zhang
http://doi.ieeecomputersociety.org/10.1109/CVPR.2006.35	67	A New Formulation for Shape from Shading for Non-Lambertian Surfaces Abdelrehim H. Ahmed and Aly A. Farag Computer Vision and Image Processing Laboratory (CVIP) University of Louisville, Louisville, KY, 40292 {abdo,farag}@cvip.uofl.edu	acknowledgements addition advanced algorithm ambiguity american analysis applications approach approaches approximation assumptions bardi based basrelief been belhumeur bellman between beyond birkhuser black camilli capuzzo computer computing concave conference control convex crandall cryer cvpr czech data developed diego diffuse discussions distance dolcetta dpvt eccv efficiency eliminates emmanuel enumath equations european evaluated experimental explicit falcone faugeras florence formulation foundation francisco friedrichs from generalization global grant greece hamilton hancock height help iciap ieee illustrate image implications information inria intelligence international ischia italy jacobi josa journal june kriegman lambert lambertian light like linear lions local lumination machine mathematical mathematics method model modeling national nayar need normals numerical object obtained optimal oren pattern pentland performance points porto posed prados prague problem proceedings processing promising proposed ragheb real recognition references reflectance reflection removing republic research results rough roughened rouy sagona scheme science second seeing seghini sets several shading shadow shah shape siam singular society solution solutions solve source sparrow specular stockholm successfully supported surface surfaces survey sweden sweeping symposium synthetic thank theory this torrance tourin trans transmission tsai unifying unrealistic using valuable viscosity vision visualization well wellposed with without would york yuille zhang
http://doi.ieeecomputersociety.org/10.1109/CVPR.2006.297	46	Successive Convex Matching for Action Detection Hao Jiang, Mark S. Drew and Ze-Nian Li School of Computing Science, Simon Fraser University, Burnaby BC, Canada V5A 1S6	action actions algorithm analysis approach approximate arbitrary articulated baker ball baseball based behavior belief berg besag body boykov capture caputo carlsson cheung chvatal combining computer conference configurations context correlation correspondence cuts cvpr demos detail dirty distance distortion drew eccv efficient efros emering energy estimating estimation exemplars fast felzenszwalb figure finding forsyth frame frames freeman fromsilhouette gesture graph graphs handbook herbelin html http human humans huttenlocher iccv icpr ieee information interactive irani jiang kanade kidsroom kinematics laptev learning linear lncs local lond malik matches matching media minimization models mori motion narrative object objects optimality optimizing pami parse pattern people pictorial pictures playspace pose poses preserving proceedings product programming propagation ramanan recognition recognizing recovering references response ronfard schmid schuldt searching segmentation sequence shape shechtman shortlist society solutions space statis statistical strike structures stylized sullivan templates theory throwing time tracking trans triggs using veksler versus virtual vision vismod weiss wiley with workshop york zabih zisserman
http://doi.ieeecomputersociety.org/10.1109/CVPR.2006.174	44	Learning Temporal Sequence Model from Partially Labeled Data	aaai ability abnormality action actions activities activity additive agent aggarwal algorithms analysis annals asymmetrically automatically average basri bayesian berg berkeley blank bobick body boost boosted boosting calibration channels chellappa choudhury classification common computer conference confidence correct correctly cviu cvpr data dataset detecting detection discovering distance division dropit duong dynamic efros equal essa event exercise frames france friedman from garg glucose gong gorelick group hakeem hastie hidden hongeng horvitz huang human iccv icme icmi ieee image improved individual indoor inference inferring insert interactions invariants irani ivanov july june kautz labeled layered learning liao loboa logistic lukowicz machine malik markctrl markov mined minnen models mori motion multi multiple murphy negative networks nevatia nice node nodes number office oliver ordered over overall pages pami parameswaran parsing partially pattern philipose phung pnet positive predictions pressc probabilistic proceedings propagation rated readbottle readid reading readresult recognition recognizing references regression rehg representation representations result review routines science screw semi sense sensor sensory sequential shah shakeit shapes shapire shechtman singer size smith society space speech starner statistical statistics stochastic success switching table temporal thesis tibshirani time training transportation troester turnoff turnon understanding unscrew unsupervised unusual using venkatesh video videos view vision visontai visual vitoria wait ward washington with wyatt xiang zhong
http://doi.ieeecomputersociety.org/10.1109/CVPR.2006.223	31	Person Reidentification Using Spatiotemporal Appearance	accurate achieved addition addresses aerial affine algorithm ambient amit anal analysis analysisbased animation appearance application applications applied approach approaches articulated articulation assembly associated automatic ballard based been belongie bhattacharjee bookstein both calibration camera cameras candidate canny capturing cases clothing collins color colour combined comparison computational computer computing conclusion conference confirms confound consistent content contexts correspondence correspondences costume crosscorrelation cvpr darpa data database decomposition deformable demonstrated detection detectors developed digital dimensions directions directly duff dynamic edge efficient enabling establishing estimation european evaluate even experiments extended face fast feature features felzenszwalb finally first fitting flow following forensic form forth from function gait generates generating generation goncalves google gool graphbased graphical graphics greatly hendriks hierarchical highly histograms human huttenlocher hypothesis identification ieee image imagery images immersion improved indexing indicate information intell intelligence inter interest interface international introduction invariant involving issues jaffre jain javed jojic joly journal kadir kang kong kunt labeling lagendijk landmark large larger learning local lowe mach machine malik matas matches matching matrix merging method mikolajczyk model modeling moscheni motion mottaleb much multiple nevatia ning novel number numbers object observe operator optical ordered over overlapping page pages paper part patras pattern people periods perona person plan point polar porikli presented probabilistic proceeding proceedings processing provides puzicha queries recognition reducing references region registration reidentification representation resolution retrieval riao robust scale schaffalitzky schemes schmid science second segmentation segments seigneur senior september shafique shah shape shapes shevlin shoemake signatures silhouette simulations sivic size society soille solis song spaces spatiotemporal springer stage statistical structural subjects swain templates text that this though through time tracking trans transactions triangulated tuytelaars understanding unsupervised used using valuable video videos vincent vision wang watershed watersheds will with work workshop would zhang zhao zisserman zitnick
http://doi.ieeecomputersociety.org/10.1109/CVPR.2006.324	40	Using dependent regions for object categorization in a generative framework	alexander algorithm allocation american among analysis appearance association bags bayesian beal berg best beyond blei boosting brady caltech categories category class classes classification collections collins combining computer conclusion conditional conf conference construct conventional correspondences cortex csail cvpr darrell dataset datasets date david degrees dependence dependency describing description detection different dimensional dirichlet discloses discovering discriminative distortion distribution each efficient efros european explore extended extends features fergus fields find fisher framework freeman from fruitful future generative grauman handle hierarchical hierarchy hmax holub ieee ijcv image information inspired intelligence international invariant jitendra jordan journal june kadir kernels kristen language larger latent lazebnik learning leung local localized lowe machine maire malik match matching materials michael model modeling models multiclass murphy mutch natural nearest neighbor nips novel object objects pages paper patches pattern performances perona poggio ponce press proc procedures proceedings processes propose pyramid quattoni random recognition recognizing references relevancy report reported representing research retrieval russell saliency scale scene scenes schmid semantic serre sets shape sharing shot sigir similarity sivic society sparse spatial statistical submitted such sudderth system taxonomy technical test textons that theme therefore these this three torralba transactions transformed trevor unsupervised using utilizing variations version vision visual volume weber welling which will willsky with wolf words zhang zisserman
http://doi.ieeecomputersociety.org/10.1109/CVPR.2006.33	76	A Multi-Channel Algorithm for Edge Detection Under Varying Lighting Conditions	accurate acknowledgement acquisition algorithm almageed also analysis application applications applied approach associated authors average been behaviour berkeley better bilmes black bonnin borzenko both build california camera canada canadian canny cases changes channel characteristic chinnaswamy chopra cito color combinations communication complete composed computation computational computed computer conditions conf conference consistent controlled controlling cooperative corner corresponds crucial cumani currently cvgip cvpr cybern data debrunner deployment depth described detecting detection detector discards docking douarin dramatically dynamic edge edges estimation experiments exposure extreme fail feris financial fixture flash flow frequently from fusion gaussian generally gentle gradient graphical graphics grappling hancock hardware help hidden high hoeltzener however humans icip icpr icsi icvs ieee illumination image imagery images imaging implementation improved including increases indicates individual influence information input intel intelligent intelligently interpretation intl jane jasiobedzki jenkin jepson lesperance less lights like limitations linux logical lopez ludtke majority manuscript mapping markov master medical memory merging method mining minute missions mixture model models mulligan multi multiple multispectral mutli need needed noise noisy nonphotorealistic number obsniuk occupy only ontario operate optical orientation outliers outperforms over page pages pami paper parallel parallelization parameter part particular parts pattern performance pissaloux pixels pixelwise pollution porikli pose preparing pressure proc proceedings process processing project proposed prototype random range raskar recognition reduced references remote rendering report requires research resolution response responses robot robust ruiz running same samples sampling scene seconds segmentation segments sensing sensor separately server shadow show since size small smith snakes society some southwest space spacecraft special spectral speed stable statistical statistically structure stylized such suen support symp syst system systems tains technical technology teruel thank that thesis this those toronto tracking trans turk tutorial twenty under underground underlying underwater university unstable used using varying very vision volume voting which whole will wilson window with would xeon york
http://doi.ieeecomputersociety.org/10.1109/CVPR.2006.49	51	AdaBoost.MRF: Boosted Markov Random Forests and Application to Multilevel Activity Recognition Tran The Truyen , Dinh Q. Phung , Hung H. Bui , Svetha Venkatesh 	abstract activities activity adaboost addition advance advances algorithm algorithms also analysis appears applications applied apply approximately arbitrary artificial ashenfelter aspects athena automatically banff bartlett baxter behaviour belmont bertsekas besag boosted boosting bottom bounds breaking bulatov cambridge canada class classification classifiers combining compared computer conclusion conditional conf conference confidence contextual crfs cvpr data dcrfs detecting detection diego dietterich discriminative discriminatively discussions distributions dynamic each edition efficient estimate estimation exhibits experiments factorized factors fields figure fourth framework frean freeman frequently from function functional furthermore garg general generatively giving gradient ground hardy have hebert heskes hidden hierarchical higher horvitz however human hypotheses iccv icml ieee important improved inequalities information intelligence interaction interfaces into intractability investigate issue jaakkola joint journal kanaujia knowing kumar labeling lafferty large lattice layered learn learning left level likelihood limited littlewood logarithmic machine margin markov mason massachussets maximum mccallum metaxas method model models more motion movement multilevel multimodal murphy networks nguyen nips nonlinear novel object offers oliver only opinion optimal original others pages parameter partition pattern pereira phung pick plan policy pools predictions presented press primitive probabilistic problem proc proceedings programming proposed random range rated recently recognition recover references representations research rohanimanesh round royal schapire scientific segmentation segmenting select selecting selectively sequence series should since singer sminchisescu society some spatial statistical stress structure sutton systems tackle techniques than that them theory these those times torralba tractable training trajectories trans tree trees truth unanswered university upper using variables venkatesh vision volume wainwright weighting weights west wider willsky with without
http://doi.ieeecomputersociety.org/10.1109/CVPR.2006.241	65	Reciprocal Image Features for Uncalibrated Helmholtz Stereopsis Todd Zickler Division of Engineering and Applied Sciences, Harvard University	accurate acknowledgment affine analysis arbitrary attention auto automatic available award baker based baseline belhumeur beyond binocular brand brdfs british calibration camera career carnegie case chum code comparison computer conf conference content context cvpr data datadriven demand dense descriptors detection detectors developed development dimensional distinctive drbohlav estimating european evaluation exploiting extremal feature features fill flexible foundation framework from fundamental funded gool graphics gross growing guillemaut helmholtz here however ieee illingworth image implementation important institute intelligence interest invariant janko journal june kadir kanade keypoints kriegman lambert largely likely lindeberg local lowe lucas machine magda making matas matching matrices matrix matthews matusik maximally mcmillan mellon mendonc methods mikolajczyk model murray must national need order pages pair pajdla part past pattern performance pfister play point ponce practical presented proc proceedings processing projective radiometric ransac reciprocity reciprocityspecific recognition reconstructing reconstruction references reflectance region regions remains report ritr robotics robust role rough scale schaffalitzky schmid science selection self separated sift single society some stable stereo stereopsis stratification strongly such surface surfaces symp systems tech technique textured thank there this those three tools torr toward trans transactions transmission tuytelaars unexplored unifying univ urban views vision visual visualization volume wide widely will with work years zhang zickler zisserman
http://doi.ieeecomputersociety.org/10.1109/CVPR.2006.132	30	Hidden Conditional Random Fields for Gesture Recognition Sy Bor Wang Ariadna Quattoni Louis-Philippe Morency David Demirdjian Trevor Darrell	aaai ability acero action adaptive advantages american amount appearance application applications approach approaches articulated articulation assan assigned automatic based below between boosted both brand callum certain class classes classification clear collins combines communication complex computer conclusion conditional conf conference conll consecutive constrained contextual conversation coreference corresponding coupled crfs culotta current cvpr darrell data deictic demirdjian demonstrating dependencies dependency detection detector discriminative displayed each early ejiri enhanced entity entropy even extraction face feature fields figure first four frame frames framework freeman freitag from fujie gest gesture gestures glass groebel gunawardana hand have hcrf hcrfs head herbert hidden hmms huang human iccv icmi icml identity ieee ijcai important improved incorporating increasing induction inference information integration intelligent interaction interactive interfaces international interpretation interspeech iscv joint kanaujia kapoor kaufmann kobayashi kumar label labelling lafferty lang language latent learning lexicons linguistic livescu long loosely mahajan main markov matlab matsusaka maximum mccallum metaxas model models morency morgan motion multi multimodal murphy nakajima named networks newton nips nition noun november object oliver optimization outperform outperforms pages pami para pattern pavlovic pearl pentland pereira performance perspective phone picard platt plausible pose presented probabilistic proc proceedings proper proposed quasi quattoni rabiner rahimi random range real reasoning recog recognition reference references regarding results robot saenko segmentation segmenting selected september sequence shake share sharma shown sidner sign similar since single siracusa size sizes sminchisescu society some speech starner state states streams structure synchronized systems table tasks that them these this time toolbox torralba toward tracking train tutorial uncertainty uniquely untethered used user using variable varying video viewbased visibly vision visual volume wellner when while wilson window with wkshp wksp work workshop
http://doi.ieeecomputersociety.org/10.1109/CVPR.2006.182	68	Mesostructure from Specularity	aanaes acknowledgements acquiring acquisition adelson after against algorithms also analysis anderson anisotropic anonymous another anrys application applications applied approach appying apricot arbitrary artifacts assistance based because below bernardini beyond bidirectional bilateral black blinn bouguet bouguetj boult brdf brdfs btfmapped buelthoff build building bump calib calibration caltech camera cameras candy capture captured capturing chang checkerboard chellappa chen chocolate christian circularshaped classification claus close cognitive color combining comments comp compactness compare compares complete complex components computer conference control covered covering cues curless cviu cvpr dana dark data davis deformations demonstrated dense depth detailed details detected determining developed difficulties diffuse digital dimensional direction doorn dpvt dried dutre eccv effectively effectiveness efficiently egrw egwr enforc engineering example expected exploit extended extension extensive fang favorably field figure filtering final fine fitzgibbon flash flection flections fleming florentine form forum four framework frankot free from fuchs fulk functions further future general geometric geometry ginneken ginsberg ginzton godin goldman gool graph graphics graphs grateful gray grooves ground gueziec hairs handle help hertzmann heyden highlight highlights highly homogeneous http human hybrid iccv ieee ijcv ikeuchi illuminance illumination image images improve including incrementally information inhomogeneous input inside insightful inspection integrability interactively international intrinsic jelly journal kanade kappers klein klinker koenderink koller kutulakos lambert lambertian large larson laser last layer leather level levoy light lighting like local magda main make manduchi martin masselus material materials matlab measuring meseth mesostructure method michelangelo minolta mirror missed mittleman model modeling more motion muller multiple murata nayar nehab neubeck normal normals norman novel object objects observation optical orange order oren orientation other pages pami part paterson path pattern perception pereira performed perona photography photometric piece pieta pixels polarization position positions powder precise preparing proceedings project properties psychol pulli ramamoorthi range real realistically realworld rebuttal recognition reconstructed reconstructing reconstruction recovered recovering references reflectance reflection refractive regular reliable relief render rendered rendering representations research resulting reveals reviewers rushmeier rusinkiewicz samples sanderson sarlette sattler savarese scale scan scanner scanning scattered science seitz separate separating separation shade shading shadow shafer shape shows shum siggraph simple simulation single skin snyder society solem some source sources spacetime sparse spatially specular specularities specularity sphere spheres stage statistical statues steger stereo strong structure structured successfully surface surfaces synthesis synthesizing system tang tangent taubin texture textured thank thanks that their then theory this three threshold todd tomasi tong toolbox torralba trans transactions translucent trends triangulation truth types umeyama under unifying unreliable used user using variation variational variety varying very viewpoints vision visual wang weiss when which white with wolff woodham workshop world would wrinkled yields zalesny zheng zickler
http://doi.ieeecomputersociety.org/10.1109/CVPR.2006.194	116	Multi-Camera Scene Flow by Tracking 3-D Points and Surfels	affordable against algorithm already also alvey application artificial attempts automatic baker based become binocular bouguet breckenridge camera cameras capable capture carceroni changes collins combined comp complete completely complicated computation computationally compute computer computes computing conf conference corner corp cvpr czech data deformable dellaert derivatives detecting detector devernay diego differential difficult dimensional disappear dynamic easily eccv edge elements estimation extended extension extracted fact faugeras feature features field filtering flow footprint framework from geometry give good greece ground handling harris have head hilton huber iccv ieee ijcv illumination image images implementation integrated intel intelligence intelligent international involved iterative john joint june kambhamettu kanade keriven kerkyra keypoints kutulakos labs lead least lies little lucas lucaskanade main make matthews measurements methods microprocessor modelling models more most motion much multi multiple multiview necessary need notably number objects occluded once optical optimization orientation other pages particle parts patches pattern pixels planar points pons posed position prague press primitives probably problem problems proc proceedings process properties propose proposed pyramidal rander recognition reconstruct reconstructing reconstruction recovery references reflectance registering registration rely report represent representation republic research resolved restriction results rigid robotic robust sampling satisfactory scale scene scenes sclaroff seattle sequences several shape shapes simple small society sons squares stable standard statistics stephens stereo stereoscopic still streams structure studied such super surface surfel surfels systems technical technique techniques texture than that their there these they thorpe three through thrun tomasi track tracker tracking tracks triggs truth under unifying used validated validation vedula very video view visibility vision wacv whereas which wiley will with without work years york zhang
http://doi.ieeecomputersociety.org/10.1109/CVPR.2006.262	101	Satellite Features for the Classification of Visually Similar Classes	about achieved acknowledgments across actresses added advances advantage again agarwal alarms algorithm algorithms alternative anchor anchors another appearance appearances assumes assumption automatic averaged awan barrymore bart based because between binary blue bmvc bottom brunelli byvatov called cameron cannot case categorization charlize chosen class classes classification classifier clearly codebooks collected columns combining comparison compensate complex complexity comprised computer conditional conducted conf conference constellation constrain control correspond corresponding crandall creating crossvalidated crucial current curve curves cvpr darpa database dataset datasets described detect detected detection diaz difference differences different difficult direction directions discover discriminative discussion distinguishing distortions drew each eccv edge efficient epshtein equivalent error even example examples existing experiment extended extending extension extracted face faces fact false fans fast feature features felzenszwalb female females fergus figure fixed fleuret fragments frags framework from frontal function future gender gesture golomb grant grimson gwyneth heisele hits holub human huttenlocher hyperbf iccv identification identifies identifying ieee image images implementation include incorporating information informative inhomogeneous interleaved intermediate intl invariant journal jurie laboratoryfor large lawrence learning left leibe local location locations machine machines male mean method methods modeling models moghaddam more moross most motor must mutual naquet network networks neural neurosci nips note novel number object objects often other over overlayed paltrow paper part parts pattern performance perona person pixels poggio pontil presented previous priors probabilistic proc proceedings processing proposed randomly rate recognition recognizing references relative relaxed reliably reliablydetected representation representative representing research respect rest resulting results rich right roth roughly rows runs sali same sample satellite satellitebased satellites scale schemes schiele sedan sedans segmentation sejnowski selected selection semantically serre sexnet show shown shows significant similar size socalled society sparse spatial specific statistical statistically studying subclasses substantial such support supported systems taxi taxis test that their there theron these they this thurman time tpami training triggs type ullman unable understanding uninformative unsupervised used useful uses using vans variability vector vehicle vehicles vetter vidal view viewing vision visual visually well were when where with without work workshop yang zisserman
http://doi.ieeecomputersociety.org/10.1109/CVPR.2006.39	130	A Refractive Camera for Acquiring Stereo and Super-resolution Images Chunyu Gao and Narendra Ahuja Department of Electrical and Computer Engineering University of Illinois at Urbana-Champaign Email: cgao@uiuc.edu, ahuja@vision.ai.uiuc.edu	acquisition adelson advances aggarwal ahuja algorithm alternating analysis applications applied april assaf automation based baseline biemond bilinear biprism blurred bombay borman boston cambridge camera chae chun chunyu circuits cliffs comparison comprehensive compressed computer conference control correspondence cvpr cybern dame daniel data december dense depth detector dhond different digital directions disparities distance down edition edward elad engelhardt engineering englewood enhancement epipolar evaluation ezra feature february feuer field flow focus formulation frame from fundamentals future generalized greenwich ground hall hausler heesung high hill huang hyongsuk hyun icpr identification ieee image images india industrial intelligence international interpolation iterative jaehong jain january jitter john journal kanade katsaggelos kluwer kweon lagendijk lens lines machine magazine maximum mcgraw measured measurement method methods mirror mirrors model modern molina moshe multiple narendra nayar nearest neighbor nene nishimoto noisy notre novel observations okutomi optical optics orthogonal pami parallel pattern pentland planar plate plenoptic prentice proc proceedings processing projections proposed qualitative recognition references registration report research resolution resolved restoration review richard robotics rotating sampled scharstein segall seiken sense sensing sequence sequences several shin shirai shree signal single small smith society song spatial stereo stevenson structure super superresolution symposium system systems szeliski taken taxonomy tech through tokyo trans transactions transs truth tsai univ using video vision volume wang warren with without workshop youla zomet zoom
http://doi.ieeecomputersociety.org/10.1109/CVPR.2006.67	63	Automatic Landmark Tracking and its Application to the Optimization of Brain Conformal Mapping	academic across algorithm align also analyses analysis anatomical application approach atlases automatic averaging based bottom brain cambridge cerebral challenges chan chapter cipolla computational computer conference conformal consistently coordinate corbetta cortex cortical cramon creating curvatures curves cvpr dale data davatzikos deformation derivatives detection diffeomorphisms done drury duncan essen estimating examined experimental explicit extract features feautures finding fischl framework from future genus geom giblin glaune global harmonic hayashi high huang human hybrid iccv ieee image images imaging implicit important international intersubject ipmi journal kruggel landmark large lectures leow level lines lohmann manifolds mapping maps match matching mathematical maths mega meshes method miccai miller model more motion multiphase mumford neuroimage nicolson novel numerical outer pages palm parameterization parametrization pattern pdes population press prince proceedings processing protas quantitative recognition references resolution results ribbon riemann rusinkiewicz schoen schultz segmentation sereno shah shape show snyder society solving some sphere springs staib statistical structural structure subjects sulcal surface surfaces symp system that their thompson toga tootell trans triangle university using vaillant vese vision visual vlsm volume wang warping well which will woods zeng zero
http://doi.ieeecomputersociety.org/10.1109/CVPR.2006.118	6	Fast Compact City Modeling for Navigation Pre-Visualization Nico Cornelis, Kurt Cornelis and Luc Van Gool Katholieke Universiteit Leuven Kasteelpark Arenberg 10 B-3001 Heverlee, Belgium	abidi accuracy accurate achieved acknowledge acknowledgements adapted additional aerial airborne algorithm algorithms allen applications approaches april area artifacts assumptions automatic automation based because borshukov brenner building buildings builds calculation cambridge camera cameras capture card carving cities city clever clouds commission commodity compact completely computation computer conclusion conference connectivity constrained constructing construction consumer content cornelis correspondence corresponding covering cvpr data databases debevec dense dependent depth detailed determine determining devernay different dijkman dirac distribution easy editor efficient entire eurographics european evaluation extracted extraction facade fast feature february figure final focus frame framework fritsch from frueh fund further future generating generation geometry gool graphics ground gruen haala hand hardware hartley high hobbie iaprs icpr ieee ijcv image images implementation incorporates increase indoor integral integrated international isprs jain june kauai kept koschan large laser latter leuven level livers maas made mainly making mapped mapping measurements merging meshes methods metric mobile model modeling models monochrome more moreau mounting multi multiple navigation needed neumann never object objects october onto optimized other outdoor output pages paik pattern photogram photogrammetric photographs plans platform point pollefeys poses precalibrated presented press problems proc proceedings processing production programmable project projective provider range real realism realistic reality recognition reconstruction references remove rendering representations required research resolution resulting results retrieved robust scale scanner scans scene scenes scharstein separate sequence signals silhouette simple simplified situations society space speed stamos starting statter stereo stuttgart subsequently suitability support survey system szeliski taxonomy teleatlas texture textured texturemapping textures thanks their third this time tools topological traffic true university unleashed urban used uses using variable vehicles veksler vestri video view viewdependent virtual vision visualization volume vosselman voted week when which will window wish with wolf work workshop xxxiv yang zakhor zisserman
http://doi.ieeecomputersociety.org/10.1109/CVPR.2006.237	27	Real-time Hand Pose Recognition Using Low-Resolution Depth Images Zhenyao Mo, Ulrich Neumann University of Southern California Computer Graphics and Immersive Technologies Lab {zmo, uneuman}@graphics.usc.edu	adapted adopting algorithm algorithms ambiguities ambiguity analysis angles appearance application approach approximate articulated articulation athitsos augmented automatic average based basic because benefit blur both build camera candidates canesta cannot capturing cipolla classification cohen color comparison compelling computer conclusion conf conference constraints consuming correctly crowley cvpr department depth desk different difficult digit discrete dynamic eccv efficiency efficient electronic encoded encoding estimates estimation european example extend face failure figure finger fingerspelling fingertips framework from frontal gesture gestures hand high however http huang human iccv iciap icpr icvs ieee image images impact implementation incorporating independent inequality input instead intelligence intensity interaction interface interfaces interpretation intl joint july kanade kansas kimura koike kuno languages large laser left lexical limitation linguistics machine manual many martin matser measurement mendonca miura model monocular most motion multiple natural neumann next number only pages palm paper parameter pattern pavlovic pentium perception performed personal plausible pose poses postures present proc proceedings processor propose provide provides pursue quantitatively range ratfgrts real realtime recognition recognizable recognize recognized recognizing reconstruction reference references refinement region rehg reliable remove represent represented representing resolution retrieval review right robust rosenberg runs sato schema sclaroff segmentation shape sharma shimada shirai signed significantly signwriting simple size society solution sophisticated space spelling springer stenger steps structures suffer sufficient support system systems that thereby thesis this thus time together total tracking transaction university users using vectors verlag version view viewpoint views vision visual vocabulary volume which wide will with within workshop writing
http://doi.ieeecomputersociety.org/10.1109/CVPR.2006.65	45	Automatic Discoveryof Action Taxonomies from Multiple Views Daniel Weinland Remi Ronfard Edmond Boyer	action actions activities actors agarwal algorithms aloimonos also ambiguous analysis anandan andg animated applied august avatars azarbayejani based becker blake bobick boyer brand build bysimultaneouslylearning campbell cham classification classifiers complex component composite compositeactions computer concept concepts conference correspondences currently cvpr davis december deform description descriptions detectionusing discover discovery efficient event features february feng fermul fields free from fukunaga futurework gender gesture gestures grammars green guan have heads hebert hierarchies hierarchy history hoogs human icml ieee ijcv images independently iniccv intentionalmotion invariant irani karapurkar kettnaker kojima language learning manor marrand mathematical meaningful method modelingandrecognitionofhumanactionsusinggrammars modelling modelsof monocular more motion movement movements natural neumann october ofanimate ogale pages pami pattern patterns pentland phil primitive proc proceedings quantifying recognition recognizing references representation rittscher ronfard segmentation segmenting semi sequences sets shapes shum silhouettes small society spatio stein style styles sukthankar supervised tamura tcsv templates temporal their track training trans triggs unsupervised using vaina video view viewpoint vision visual vocabulary volumes volumetricfeatures wang weinland weplantousethosetechniquestolearnstatistical with zelnik zheng
http://doi.ieeecomputersociety.org/10.1109/CVPR.2006.168	18	Learning Exemplar-Based Categorization for the Detection of Multi-View Multi-Pose Objects Ying Shan Feng Han Harpreet S. Sawhney Rakesh Kumar Sarnoff Corporation 201 Washington Road Princeton, NJ 08540, USA	accurate acknowledgements adaboost also analysis application applying approach approaches aspects assistance attention automatic based batch better between blake boosted boosting both cameras candidate cascade cascading choosing class classes classification clustering clutter coarseto compared comparing components computer conditions conf conference customized cvpr dalal decision depending design detection detectors different discriminative discussions distance does driving during eccv edge efficacy european examine example exemplar exemplars experiments explore extend face fashion fast feasibility feature features field fine focus frame freund future gavrila gdalyahu general generalization gradients handle hashing hausdorff have hayon hebert hierarchical histograms huang huber human huttenlocher iccv ieee illumination image images impact improving initial inspiring intelligence intelligent international into intra invariant investigate jones journal june kind klanderman kohonen kumar large learnability learning level like linear look machine matching matei measures merl method metric model models mohan multi multiple needs nevatia object objects observed olson online only organizing oriented overlapping pages pami papageorgiou pattern pedestrian people performance philomin philosophies pixels poggio portion pose possibility potential probabilistic proceedings processing professor proposed ransac rapid real recognition references related report representation rotation rucklidge same sawhney scene schapire science segmentation selection self shan sharing shashua shum simple single small smart society song space statistical strated strength study support symposium system systems target technical thank that their theoretic this time ting toend toyama tracking trans tree triggs type ucla unsupervised using valuable variations vector vehicle vehicles view viola vision volume weak will with work would ying zhang zhuowen
http://doi.ieeecomputersociety.org/10.1109/CVPR.2006.335	148	Wire Structure Pattern Extraction and Tracking From X-Ray Images of composite Mechanisms D. TschumperleŽ, J. Fadili GREYC IMAGE (CNRS UMR 6072) 5 Bd MarechalŽ Juin, 14050 Caen Cedex, France	abergel acquisition algorithm analysis anisotropic appear applications applied approach april arctic aspect aubert based because block bruce cabral caen calculus caltech cambridge cancun candes cessing color combination common complete complicating composite computational computer computing conference constraints contrast convolution coordinate corfu curvature curvelet cvpr database december decomposition deep deeper demanet denoising deriche deserves detection different differential diffusion direct directed discrete donoho edge elad engineering enhance enhanced enhancement ensi equations export factors fast fiber fields filamentary finally flannery forth fortran framework geometric gessat goal graphical graphics gray harmonic hawaii houston ieee ijcv image images imaging integral intelligence international isolate january journal july june kauai kornprobst kutta large leedom line machine malik many marcoux marriott master materials mathematical mechanics method methods mexico minimization modeling much multiscale multivalued murtagh nonetheless nonparametric numerical observed obtained october odru offshore omae ongoing onto operations original oscillating osher other output partial pattern patterns perona pipeline plane poirette possible precise preliminary preservation preserving press prior problem problems proc proceedings processing project projection proposed quality recipes recognition references regularization relaxation report representations results riser runge saint sample sapiro sardy scale sciences scientific separate simplified smoothing society space sparse springer stage starck stassen statistics step still structure structures stuttgart successfully system systems technical tensor tensorbased teubner teukolsky textures then thesis this time tools total toussaint towards tracking trans transactions transform transforms tschumperle tseng ultra university used using validated valued variation variational variations vector verlag very vese vetterling vision wavelet weickert were while with work ying
http://doi.ieeecomputersociety.org/10.1109/CVPR.2006.154	43	Inferring Facial Action Units with Causal Relations Yan Tong, Wenhui Liao, Qiang Ji Department of Electrical, Computer, and Systems Engineering Rensselaer Polytechnic Institute, Troy, NY 12180-3590, USA {tongy2, liaow, jiq}@rpi.edu	action actions active actively activities akamatsu alberta alto analysis annals application applications applied approach artificial asymmetric automatic autonomous banff bartlett based bayesian behavior between canada canadian chen classification classifying coding cohen cohn comm comparison complexity comprehensive computations computer conf conference consulting cootes cvpr cvprw cybernetics data database davis detection dimension donato dynamic dynamics ekman essa estimating evaluation event evidence experiments expression expressions extended face faces facial fasel february flexible flow frank friesen from fully fusion gabor gaborwavelet garg geometry gestures hager heckerman huang human icpr ieee image images improvements increasing inferring information integration intelligence intensities interaction international interpretation into jones journal june kanade kapoor lainscsek lanitis layer learned learning lien littlewort long luettin lyons machine machines mamdani march measurement measurements microsoft model modeling models more movellan movement multi network networkg networks october optical oriented pages palo pami pantic part patras pattern pentland perceptron picard point poole press probabilistic proc proceedings psychologists real recognition recognize recognizing references relationships report representation robotics robust rothkrantz schuster schwarz search sebe sejnowski sequences show significant simple society spontaneous static statistics support surements system task taylor technical technique temporal tenth then tian time tracked tracking trans tutorial understanding unit units upper using valstar various vector video viola vision wavelets well wellman will with workshop world yacoob yields zhang
http://doi.ieeecomputersociety.org/10.1109/CVPR.2006.176	59	Local Features, All Grown Up Andrea Vedaldi Stefano Soatto University of California at Los Angeles Los Angeles ­ CA 90095	accidental accidentally acknowledgments acoustics addressed affine afosr agarwal algorithms aligment allan analysis appearance approach areas artificelle automatic available background baker based bayes bayesian belhumeur belongie better black blake bmvc bookstein brady building calculus caselles categories category certain changes chemestry chen cipolla clutter coll comparison competition complex complicated computational computer conf conference constraint contains contour contrast control cootes correlated cremers criteria criterion cvpr decomposition deformable deformation deformations dense derivatives descriptors design detecting detection detectors diffeomorphic differential direct directly distinctive easy eccv editors efficiency efficient engineering enough ensuring equivalence especially evaluation ever example examples exhaustive expensive fact fast faugeras faugerasn feature features felzenszwalb fergus ferrari finding fleet flow formes forsyth framework from fuctions gaussian generative geometry gesu global globally gool grouping groupwise growing grows haddon hager handbook have hermosillo huttenlocher iccv ieee ijcv ilic illumination image images imposed incorporates incremental induced information integrals intelligence interest international invariance invariant ioffe issue jepson journal kadir kanade karl keypoints kikinis kinghorn latter lazebnik learning level lhuillier local lowe lucas maps marsland matas match matching mathematical matrix matthews measures medical method might mikolajczyk model modeling models monocular more morel motion multiband multiple mundy natural nonrigid novel object objects occlusions optimal optimize orientation pami paragios parameterization parametric parametrized part pattern pdes performance perona physically pichon pictorial piecewise planar plate ponce portions position primitives principal problem proc proceedings processing progressive properly quan quantum recognition reconnaisance references region registration regularization report requirement research rigid roadmap rothganger saliency salient salzmann same scale scenes schaffalitzky schmid schmidt segmentation selection shape signal significantly similar simultaneous since single singularities smith snakes soatto society solution solving sparse speech spline splines sponsored springer sprites stable statistical statistically statistics stereo structure structures substitute surface tannenbaum task taylor technical term tested textured than that theoretical these thin this thus titsias topographic toward tracking training trajectories trans triggs true tsitsiklis tuytelaars twining uncluttered under uniform unifying unoccluded unseen unsupervised using valid variational vedaldi verlag view viewpoint views vision visual warps went what where which while williams wills with without workshop yacoob yang years yuille zisserman
http://doi.ieeecomputersociety.org/10.1109/CVPR.2006.145	85	Image Pre-conditioning for Out-of-Focus Projector Blur Michael S. Brown Peng Song Tat-Jen Cham School of Computer Engineering Nanyang Technological University, Singapore 639798 msbrown@ntu.edu.sg psong@pmail.ntu.edu.sg astjcham@ntu.edu.sg	acceptable accommodation achieving across addison aggarwal ahuja algorithm aligned alignment allowed already also analysis applied approach approaches approximation array ashdown augmented autocalibration automatic axis back bandyopadhyay barcelona based basis been belhumeur best better between blur blurring boston bottom boult branzoi brown calibrating calibration camera cameras cannot capabilities case casually cham chen city color compensation completely compositied comprehensive computer condition conditioned conditioning conference constructed contrast convolution coordinate counter creation cutts cvpr deal deeper defocus deguchi demonstrate demonstrated depth detection diego difference difficult digital discussed discussion display displayed displaying displays domain dynamic edition effectively effectiveness effects elimination enhancement equivalent essentially estimate estimated estimation examined example experiences exploiting extreme favaro features field figure filter filtering filters focus focused form formal format frame france francisco frequency from fuchs further future gaussian generating geometric geometry gill gonzalez gotz graphics grossberg hardware have hensley homography however human iccv icpr ieee image imagery images immersive imperfections implementing initially input inset intensity interactive international into inverse investigation isar issues itself jaynes lake lamps large lessen lessening light likewise limitations madison majumder many march marked methods micromirror model modeling more mosaics mullin multi multiple multiprojector nayar needed nice nonuniformity note noticeable novel objects obtain obtaining occluder office okatani only original orlando outdoor output outside over overlapping pages painting pami paper pattern performing peri personal photometric pixel pixelflex planar point possible practice preconditioned presentations presented preserved proc procams procedure proceedings process processing produce profile programmable projected projection projector projectorcamera projectors projectorscreen provide providing psfs psnr quality quantitative quite radiometric raij range raskar real reality reasonable reasons recognition reconfigurable rectification rectified references regions registered rehg related removal remove requirement research restoration result results ringing robinson salt scalable scale scene screen seales seamless second self selfcalibration sensitivity several shader shadow shadows shape sharp sharpening siggraph smarter soatto society solely solutions song spatial spatially specific specified standford steele stesin stevens stockton subjective successful sukthankar summary suppression surati surface surfaces symposium system systematic systems technique techniques technology tenenbaum texture textures that then theory these thesis this tocamera together towels towles trans trees unified uniformity used uses using values vancouver varying very vision visualization wallace warrants washington ways webb welch welcomed were wesley where which while wiener will with woods work yang
http://doi.ieeecomputersociety.org/10.1109/CVPR.2006.281	7	Single View Reconstruction of Curved Surfaces	algorithm alleviate also alternatively amount animation anjyo approximating arai assignment automatic banana based best better blake bmvc case chamfer clumsy column columns complete computational computer conference constraints contour controllable criminisi curvature curvaturelike cuts cvpr darrell dense devise difficulty dugas each early eccv edges efros energy estimating eurographics example extraction fast faugeras figure finally find fitzgibbon foreground form found free from front geometry gortler grabcut graph grauman grimson hebert highlights hoiem hope hoppe horry hulls human iccv ieee ijcv image images increases inference inflation initial input intensity interactive interface into iterated iterative kass keriven koenderink kolmogorov kumar level make match maybank mechanism mesh method methods metrology minimization minimizing model modeling modelling models more much objcut object objects oblique orange pages papers parameter pattern perturbing phocion photo picture piecewise pixel planar prasad press prior problems proc proceedings processing product realistic recognition reconstruction references reid required respectively romdhani rother samson scenes seeking segmentaion segmentation seitz shakhnarovich shape short show shows siggraph silhouette simple single skill society solid solve somewhat space specular spidery statistical stereovision strategy study sturm surfaces symmetry system terzopoulos texture than these this torr tour trans user using vetter video view views virtual vision visual ways with witkin workshop yields york zhang zisserman
http://doi.ieeecomputersociety.org/10.1109/CVPR.2006.243	8	Recognize High Resolution Faces: From Macrocosm to Microcosm	active affine analysis appearance approach avbpa banks based belhumeur chellappa chen class classes classification clausi cognitive components computer conf conference cootes cvpr database deng descriptors dimensional discriminant discriminative dual edwards eigenfaces entropy evaluation extended face features felzenszwalb fergus fields filter filters fisher fisherfaces frame framework from gaussian hespanha holub huang huttenlocher iccv ieee ijcv image images invariant jojic kernel kittler kriegman krishnaswamy lazebnik learning leung liao linear local locus luettin maitre malic matas materials maximum messer mikolajczyk modeling modelling models moghaddam mumford necessary neuro object pami pattern pentland performance perona pictorial ponce principal probabilistic problem proc proceedings projection random recognition recognizing references regions representation representing rotation sample scale schmid segmentation single size small society solve space sparse specific statistical structures subspace system tang taylor textons texture theory three towards trans turk unified unsupervised using varma vision visual vtsdb wang which winn with zhao zisserman