http://www.informatik.uni-trier.de/~ley/db/conf/cvpr/cvpr2004-1.html
CVPR 2004 (vol 1)
http://csdl.computer.org/comp/proceedings/cvpr/2004/2158/01/215810588abs.htm	78	Estimation, Smoothing, and Characterization of Apparent Diffusion Coefficient Profiles from High Angular Resolution DWI Y.Chen1, W.Guo1, Q.Zeng1, X.Yan1, F.Huang1, H.Zhang1, G. He2, B.C.Vemuri3, Y.Liu2 University of Florida, Dept. of Mathematics, Dept. of Psychiatry and Neuroscience	alexander algorithm amsterdam analysis analytical angular anisotropic anisotropy apparent appear arridge asgari attempt barker based basser belliveau biophys black blomgren brain brunberg canada cerebral chambolle chan characterization characterized chem chen chenevert chesler chileuitt classification coefficient coefficients cohen color columns complex computer conclusions conference constrained correspond cvpr data demonstration denoising denver detection determing diffusion direct echo echoes effective effects elucidated estimation evidence expression fatemi features fiber field figure first fitting form fourier framework frank from function gaussian glasgow gradient gray growth hard hasan high human identified ieee image images imaging introduced ismrm isotropic knowledge kucharczyk lazar leastsquares lebihan left level lions lower magn magnetic mapping mareci mathematik matter mattiello mcgraw measurements media method methods microstructual middle minimization minimizing mintorovitch model modeling molecular mori moseley neuronal noise nonlinear nonstandard norman novel nuclear numerische obtain obtained orientation osher parker partial particular pattern perfusion philadelphia phys physica physiological pierpaoli pipe presence principle problems proc proceedings processing profiles quantitative radiology recognition reconstruct recovery reese references regularize related removal resolution reson resonance respectively restoration right rows rudin scotland self series shapes shown simplified simultaneous simultaneously smoothing society spectra spectroscopy spin stejskal suruda system tanner techniques tensor this those three timedependent tissues total tracking tract trans transform truncated tsuruda tuch tuchand upper used using valued vancouver variation variational vector vemuri vision vivo volume voxels wang wedeen weigel weighted weinstein weisskoff weisskoffand white with workshop
http://csdl.computer.org/comp/proceedings/cvpr/2004/2158/01/215810436abs.htm	59	Programmable Imaging using a Digital Micromirror Array	academic active adaptive advances allow almost alter appearance applications applied approach apps april architecture array arrays assorted astro athale august available based become being believe bistable bloom boult branzoi bright bypixel camera castracane cause characterization christensen columbia communication comp components computer concept conclusion cone conf conference control controlled convergence coronography coyle cvpr deformable degrees dept design device devices diffraction diffractive digital direct directly discontinuities displays dudley duncan dynamic ease eccv edges effects eigenfaces emerging enables engineering euliss euro expect expected exposures eyes face february field fields figure flexible fourier from future general generation ginosar goodman gracht graph greater gutin haney have hence here hicks high highdynamic hill hilsenrath holographic hornbeck hyperspectral iccv ieee image imaged images imaging impact individual instantly intensified international into introduction intruments january journal june kang kearney learning left lens light made malbet many mask mcfadden mcgraw mems micro micromachined micromirror milojkovic minimized mirror mirrors mitsunaga models modulators multisampled murase narasimhan nayar next ninkov objects october optical optics orientations over pacific pages paper patent pattern pentland personal pixel pixels preparation press principles proc proceedings programmable projection public range recognition references reflected right same scene science segmentation sensor september shao shown significant since single slaughter smith society space spat spatial spatially spectroscopy spie structural substantially such surface syst system systems szeliski taken tech technology texas that these this through tilted time timely trans turk tyson univ using uyttendaele varying video view viewpoint vision visual volume when white wide winder with within without york zeevi
http://csdl.computer.org/comp/proceedings/cvpr/2004/2158/01/215810782abs.htm	104	Visual Tracking Using Learned Linear Subspaces	adelson analysis appearance approach articulated banerjee based belhumeur birchfield black blake brand cascia chaudhury chen color comaniciu computer condensation conditional conditions conf conference contour cvpr data decomposition density design efficient eigentracking elliptical european ference filters fleet freeman from geometry gradients graphics groups gupta hager head headtracking histograms html http huang icvgip ieee illumination image images incremental indian inference intelligence intensity isard isidoro jepson jpdaf kriegman learning lighting machine maraghi matching mean meer metric missing mittal models murase nayar nonrigid object objects online pages parametric pattern possible predictive probabilistic proc proceedings processing propagation publications ramesh real recognition references region representation robust sclaroff shift singular society space steerable time toyama tracking trans uncertain under using value values view vision visual volume what with
http://csdl.computer.org/comp/proceedings/cvpr/2004/2158/01/215810314abs.htm	42	Joint Prior Models of Neighboring Objects for 3D Image Segmentation	according achieves active algorithm also among applications around associated automated automatically balance balancing based been both boundary brain capture caselles cases chan changing comp comparing comput computer computing conclusion conf conference constrained constraining containing contour contours cootes corresponding cortical coupled cvpr data define deformable density detected deviation distances distributions duncan edges equations error errors estimate estimation euler even evolve evolving experiments faugeras figure final finding fixed focused framework function gaussian geodesic goes gray grimson haslam have hill however ieee image images improvement incorporating increases influence information initial intensities introduce ipmi joint journal july kass kimmel lagrange large lateral layer left level leventon locating locked lower manual matters mean medical method methods middle minb model models more much multi multiple mutral neighbor neighboring noise noisy note objects obtained only other over pami paper parameters parametrically patt pattern points portion positions presented prior probability proceedings process processing propagation provided range rapidly recog recognition references representation results right robustness sapiro schultz segment segmentation sensitive sets several shape shapes show shown shows simultaneously since small smallest snakes society staib standard statistical steps structure structures surfaces table taylor tempany terzopoulos test than that then three thus training transactions tsai using variance variations varied ventricle very vese virtually vision volumetric voxels wanted weights wells when white willsky with within without witkin work yang zeng
http://csdl.computer.org/comp/proceedings/cvpr/2004/2158/01/215810407abs.htm	55	Bayesian Video Matting Using Learnt Image Priors Nicholas Apostoloff and Andrew Fitzgibbon Robotics Research Group University of Oxford Oxford, OX1 4AJ, UK {nema, awf}@robots.ox.ac.uk	adelson advances agarwala alpha anandan anisotropic approach artificially avidan ayer background bayesian black blake blinn blue bottom cambridge channel chuang coarse cocke colour common complex composite compression computer computing conference consistency containing curless cvpr diffusion digital eccv effect elimination encoding estimation extract extraction figure filled fitzgibbon flow foreground framework from global graphics graylevels hannah haussecker high highlight hillman holes huang iccv ieee ijcv image images imaging input irani issue january josef journal july june layer layered layers likelihood mathematical matting maximum mixture modeling models motion motions moving multiple mumford natural note notices number occluding optical pages pattern peleg photoshop poorly press proc proceedings processing recognition reconstruction recovered references reflections renshaw representation representing resolution results robust rousso ruzon salesin sawhney scenes scharr screen second separated september sequence sequences shows siggraph sigplan simoncelli smith society solution solutions spatial spatiotemporal special springer srivastava statistical statistics subexpression subtraction suffer szeliski temporal that third tomasi tool transactions transparency transparent trimaps using verlag video vision visual volume wang wexler with zisserman
http://csdl.computer.org/comp/proceedings/cvpr/2004/2158/01/215810269abs.htm	36	Extracting Semantic Information Through Illumination Classification Aya Aner-Wolf Department of Computer Science and Applied Math The Weizmann Institute of Science Rehovot 76100, Israel E-mail: aya@weizmann.ac.il	adelson algorithms alignment america anandan arijon automatic based bayesian boundary bouthemy brockington cameraman characterization class classification color comp comparative comparison computer conference content context cvpr dept detection determining drew early eccv film filtering focal framework freeman freemna from funt gelgon giannetti grammar graphics gupta hall hastings horn house http husoy iccv ieee image images intel intrinsic irani iyengar james jones journal land language learning lighting lightness lippman machines mccann millerson models motion movies multi murphy nature nips object opencv optical oren osuna pami papageorgiou pattern pedestrian petzold picture place poggio prentice press proceedings processing randen real recognition recovering references report research retinex retrieval robust rubin santini semantic sensor sequences shading shot silman single sinha smeulders society spie springer statistical study support system taiwan tappen technical television templates texture theory time torralba understanding univ using vapnik vasconcelos vector verlag video viola vision volume wavelet work worring years young
http://csdl.computer.org/comp/proceedings/cvpr/2004/2158/01/215810582abs.htm	77	Motion Without Correspondence from Tomographic Projections by Bayesian Inversion Theory	adjustment algorithms alignment angle application applications automatic bayesian biol brandt bresler bundle cambridge carlo chapter common computer computerized computing concepts conference convergent correspondence cvpr december delaney dellaert dental determination dimensional doctoral edge editor electron engelhardt enpa estimation fessler fiducial fishman fitzgibbon flannery frank from general geometry globally guckenberger hartley heikkonen helsinki ieee image images imaging inversion james january jarv kaipio klifa koistinen kolehmainen lassas lawrence least limited lncs markers mclauchlan medical method micrographs microscope microscopy modern monte motion multiphase multiple nonlocal nucl numerical objects origin pages pattern phys pirttila plenum practice preserving press principles proc proceedings process radiographs radiology recipes recognition reconstruction references regularization regularized sauer scientific seitz series siltanen slaney society somersalo springer squares statistical struct structure synthesis technology teukolsky theorems theory thesis thorpe three thrun tilt tomographic tomography trans transmission triggs ultramicroscopy univ using vetterling view vision volume with without zisserman
http://csdl.computer.org/comp/proceedings/cvpr/2004/2158/01/215810619abs.htm	82	Selecting Ghosts and Queues from a Car Trackers Output using a Spatio-Temporal Query Language Christian Kohler ¨ Albert-Ludwig-Universitat Freiburg ¨ Institut fur Informatik ¨ Georges-Kohler-Allee ¨ 79110 Freiburg, Germany ckoehler@informatik.uni-freiburg.de	abduction about acapulco active activity advanced aiello algebras algorithm allen application architecture archive artificial august automatic based bayesian berlin bernhard binary bremond brocku buxon buxton calculus california cambridge canada castle charles chronicles city cognition cognitive cogvisys cohn communications computer conference connection constraint constraints cosc cosit cvpr declarative deriving descriptions doyle duentsch duntsch ecai eighteenth european freksa from galata gerber germany ghallab given gong habel harmelen hazarika hogg howell http icpr ieee ijcai information intelligence interaction international interpretation intervals joint journal june kauffman kaufmann knowledge ladkin lake learning line logic maddux magee maintaining massachusetts mateo mediated mexico models montanari morgan nagel nebel network networks novel november pattern presentation principles problems proceedings qualitative quebec queues randell reasoning recognition references regions relation relations relspat representation representations rich road rota scenario schreiber sciences sequences shapiro society spatial spatio starnberg surveillance swartout symbolic tarski techniques temporal textual their third thonnat towards traffic tutorial tutzing using video vision visual visually wender william
http://csdl.computer.org/comp/proceedings/cvpr/2004/2158/01/215810350abs.htm	47	Space-time Isosurface Evolution for Temporally Coherent 3D Reconstruction Bastian Goldluecke Marcus Magnor Max-Planck-Institut fur Informatik ¨ Graphics-Optics-Vision Saarbrucken, Germany ¨ {bg,magnor}@mpii.de	about account achieved active after afterwards algorithm algorithms almost analysis appear applications applied approaches arms around assessed average baker band barcelona based between beyond billboarding cambridge cameras carving caselles chop close coherence compared comparison complete computation computational computer computing concept conclusions conference consistency constant contours converged curvature cvpr data defining dependent deriche details detection different difficult dimensional each eccv edition employ energy enforces european evolution evolve fast faugeras featuring fedkiw figure final fire first flow formulated formulation frame frames framework free from fronts full functional future gbytes geodesic geometry given gives global goldluecke hamilton have hull hundred hypersurface hypersurfaces iccv idea ieee image implementing improments improve improved include informs initial instant intelligence international intersecting into intrinsic iteration iterations jacobi journal july june kanade keriven kimmel kutukalos lambert laurentini level machine madison magnor main marching mathematics memory method methods microfacet minimal minimization minimized modeling monographs more motion moving multi narrow need normal novel object objects observed obtained only operator optimal optimization optimize order osher pages paragios pattern photo photoconsistency physics plan planes posed presented press principles problem proc proceedings processes processing processors program propagating properties real reason recognition reconstruction references refined reflectance rendering representing required requires result results right running same sapiro satisfactory sbert schemes seitz september sequence sethian sets shape significant silhouette silhouettes simultaneously single slave slightly small smoothness soatto society source space spain speed springer standard static stereo summary surface surfaces switch table take takes temporal that their theory this three time towards tracking transactions twenty ultrasparc understanding unifying university until using variational vedula version very video view viewpoint vision visual volume weighted well when which while wisconsin with workshop yezzi zhao
http://csdl.computer.org/comp/proceedings/cvpr/2004/2158/01/215810158abs.htm	21	Self Shadowing and Local Illumination of Randomly Rough Surfaces Yinlong Sun Department of Computer Sciences, Purdue University, W. Lafayette, IN, 47907-1398, USA Abstract	acoustical adam addisonwesley alex animators anisotropic antennas appl applied ashikhmin backscattering based beckmann blinn brdf bristol brockelman bruce college communications comprehensive computer conducting conference corrected cvpr diffuse donald edition effect electromagnetic eugene finitely francois from generated generator geometrical ginneken graphics greenberg gregory hagfors hecht hilger ieee illumination images imperial james kajiya kenneth koenderink light lighting london massachusetts measuring michael microfacet model modeling models note ogilvy optics pattern paul peter petr phong physical pictures premoze press proc proceedings propagation radio random randomly reading realistic recognition references reflectance reflection rough roughened roughness rurfaces sancer scattering science second shadow shadowing shirley siggraph sillion simon smith society sparrow specular stavridi stogryn strauss surface surfaces synthesized theory third thomas torrance transactions tuong vision wagner ward wave waves where xiao york
http://csdl.computer.org/comp/proceedings/cvpr/2004/2158/01/215810291abs.htm	39	From Fragments to Salient Closed Boundaries: An In-Depth Study Song Wang, Jun Wang, and Toshiro Kubota Department of Computer Science and Engineering University of South Carolina, Columbia, SC 29208 songwang@cse.sc.edu, wang286@cse.sc.edu, kubota@cse.sc.edu	acknowledgements algorithm algorithms along alter american amir analysis apple approach authors background basri boundaries boundary boyer bvased cambridge carolina change chung closed closure cluttered column combinatorial comments comparison completion computation computer computing conference connected contour contours contributes convex cormen cvpr cwks data david detecting detection discrimination does efficient eigenvalues eigenvectors elder enhancement european extracted extracting feature features fields figure first fragment fragments from fund funded general generic global globallly grant graph ground grouping groups herault hill horaud ieee illusory images important improve inferring information intelligence international into introduction jacobs journal kubota leftmost leiserson like lindenbaum local locally machine mahamud many mathematical matrix mcgraw measures medioni model more most multiple natural network neural noticeable optimization organization original pages part pattern perceptual performance performs press proceedings processing providence quantitative ratio real recognition references reliably research respectively rightmost rivest robust rows salience saliency salient same sampled sarkar scholarship second segmentation setting shape shapes shashua shown shows siskind society south spectral standard stochastic structural structures synthesized systems testing texture than thank them theory this thornber transactions ullman unified university using vision wang while williams with wood work would york zucker
http://csdl.computer.org/comp/proceedings/cvpr/2004/2158/01/215810645abs.htm	86	Radiometric Alignment of Image Sequences Seon Joo Kim and Marc Pollefeys Department of Computer Science University of North Carolina Chapel Hill, NC 27599 {sjkim, marc}@cs.unc.edu Abstract 1.1 Previous Works	about accurate advantage after agathos alberta algorithms aligned allow also ambiguity annual another application applications applied approach august automatic beauchesne before being between bottom brightest calibration camera cameras cananda changes channel channels color colour combine combining comparametric compensated compute computed computer conclusion conference constancy contained correlation cpmcv cross cvpr darkest deal debevec dependent described difference different differently digital discrepancies discrepancy dynamic each eccv effect equations estimate estimated estimating estimation even experiments exponential exposed exposure exposures extend extending face figure finally find first fisher fourth france from function functions fusion future graphics grossberg have high highlights iccv ieee image images imaging improvement independently initial interest international inverse june kanade kang knowledge known left less like malik mann mapped maps method methods mitsunaga model modeling most mostly moving much multiple nayar need nice normalization normalize normalized nose note novel october only over overlapping pair paper pattern photographs picard pictures practical previous primary prior proc proceedings process processing proposed proven quantigraphic radiance radiometric ramesh range recognition recorded recovering references relighting remaining require required response resulting right sample second seen selfcalibration sequence showed shown shows siggraph society some sony space static statistical szeliski talk texture textured textures than that these third this total trans tree tsin undigital used uyttendaele video view vignetting visible vision want were what where which whole winder wisconsin with without workshop would
http://csdl.computer.org/comp/proceedings/cvpr/2004/2158/01/215810400abs.htm	54	Recovering shape and irradiance maps from rich dense texton fields Anthony Lobay and D.A. Forsyth Computer Science Division U.C. Berkeley Berkeley, CA 94720 daf@cs.berkeley.edu	able accurate affected affine again aloimonos analysis anderson animation annual another appear area arrows artificial ascribe attribute autocalibration available barr because been belongie blake both boundaries bridson case cases center change changes clerc cloth collisions competent computer computing conf conference contact contours curvature curved cvpr daley dark data david deformable deformations demonstrates dense descriptors detail detecting detection differences different differential distinctive distortion dress dresses eccv elastically elements equation estimates estimation european evaluation fall feature features fedkiw figure figures fine fleischer folds following forsyth friction friedman from furthermore garding garment good gradient graphics grayvalue ground grouping guide hastie have helpful homogeneity identify ieee illumination image images implies incorporating inference instances integration intelligence interesting introduction invariant invariants irradiance isotropy jones just keypoints lazebnik learning left leung local localizing lowe lowest machine malik mallat many maps marinos mean meaning measurements method mikolajczk mining model models mohr moments most neighborhood neighborhoods note object observable obtain offering omits orientation original overall pages particularly patches pattern performance phenomenon planar planes platt point points ponce poorer poorly prediction proc proceedings processes produces qualitatively quite reasonable recognition reconstruct reconstructed reconstruction reconstructions reconstructs recovered recovering rectified references regions relatively remain repeated representation reproduced research retrieval rich right robust rosenholtz scale scattered scene scenes schmid segmentation series shading shape should show siggraph skirt small smooth society some sparse spotted springer statistical statistics strong submitted suggesting surface surfaces synthetic terzopolous texton textons texture textured than that them theory there these they third this through tibshirani transactions treatment triggs truly truth unreliable using vere verlag view visible vision volume without witkin
http://csdl.computer.org/comp/proceedings/cvpr/2004/2158/01/215810166abs.htm	22	Radiometric Calibration of a Helmholtz Stereo Rig	accuracy accurate achieved acknowledgments after application applied arbitrary area array assumed austrian authors based before belhumeur beyond binocular brdfs calibration camera clarke close color column comp complex computer conclusions condition conex conference consistency constraint constraints correction cross curved cvpr czech data degrees discriminatory distribution dramatically eccv editions education equipped error essential evaluation experiment experimental exploiting fast following france from fund future gabay geometric grants have hayat helmholtz helmholtzian helmohltz help hungarian iccv ieee image improve improved improves including independent jaques jointly kirchhoff kriegman lambert left lighting magda matching mendonc method methods miller ministry more necessary nonuniformity numerically object objects observed optics optique otka ours pair paper parry pattern photometric physiologique plane planes points ponce posed possible power previous problem proc procedure proceedings proposed radiometric radiometrically ranges ratliff reciprocity recognition reconstructing reconstruction references reflectance reflectometry registration regularized requiring research respectively results right ross rows sceaux scene scientific scott sensors setup show showed shown shows simple single smutny snyder society spatial specially statistic stereopsis stereoscopic stratification success supported surface surfaces technology testing thank that thermodynamic this toward translation truly tune undergoing undoubtedly used using validation validity verified visible vision vladim when which will wish with without wkshp work would zickler
http://csdl.computer.org/comp/proceedings/cvpr/2004/2158/01/215810172abs.htm	23	A Variational Approach to Problems in Calibration of Multiple Cameras	accuracy accurate active adalsteinsson after alignment alpha angle appl application auto automation baker brady brown calibrating calibration cambridge camera cameras captured cipolla closed coefficients cohen coliseum color column columns completed computer conference controlled converged correction correspondences cube culbertson curvature cvpr decentering deliberate dependent devernay different distortion dotted during eccv engineering equation estimation evaluation evolution evolving experiments farag fast faugeras features figure fitzgibbon five flexible form four from fronts gelb geometry goss greyscales gurdjos hartley have heikkila herniou heyden high iccv icpr ieee image immersive implicit initialized inria interfaces intermediate intrinsic introduction iterative kang keriven known kumar labs lens lenses level line linear lines luong mach machine malzbender manipulation marszalec mask mathematical maybank melegy method methods metrology microsoft miscalibration model models mosaicing motion multi multiple murray nayar nonmetric number obtained oliensis onto orange original osher pami parameter parameters pattern pdes perceiving photogrammetric planebased point points pose poses press principles problem proc procedure proceedings projected projection projections propagating radial radiance recognition reconstructed recovering references report research respectively robotic robotics robust rows same sastry sawhney scenes segmentation selection self semiautomatic sethian shown silven simulated simultaneous single soatto sobel society solutions speed stages stein step stereo stereoscopic straight structure sturm surface swaminathan system systems tanguay tech technique teleconferencing theory theshelf three time triggs true tsai undistorted university using value vantage variational versatile view views vision weng while wide with without yezzi zhang zisserman
http://csdl.computer.org/comp/proceedings/cvpr/2004/2158/01/215810179abs.htm	24	Estimating Illumination Direction from Textured Images Manik Varma and Andrew Zisserman Dept. of Engineering Science University of Oxford, UK {manik,az}@robots.ox.ac.uk	acknowledgements addison affects albedo also ambiguity approximated aritficial because being belhumeur berry bmvc both brooks case chantler chellappa classifying cogvisys computer conf conference constant cvpr dana database direction distribution distributions eccv edition effect ellipses estimating estimation evans feiner filters financial finite foley from gaussian ginneken graphics grateful hannay hastings height heriot horn however hughes ieee ijcv illuminant integration intell involve joint just koenderink kriegman lighting lissajous math mcgunnigle mean mike must nayar number numerical pages pami pattern peacock penirschke petrou phys points practice practise principles proc proceedings project provided random real reasons recognition reconstruction reduces references reflectance relief resolved rhodes rotation schmidt scholarship shading shape sharing society source spurious statistical support surface surfaces texture texturelab textures that then theory transactions umbilic values variance vision watt well wesley which wileyinterscience with world yuille zero zheng
http://csdl.computer.org/comp/proceedings/cvpr/2004/2158/01/215810203abs.htm	27	Atlanta World: An Expectation Maximization Framework for Simultaneous Low-level Edge Grouping and Camera Calibration in Complex Man-made Environments Grant Schindler and Frank Dellaert College of Computing, Georgia Institute of Technology {grant.schindler, dellaert}@cc.gatech.edu Abstract	academic algorithm andreas antone applications automatic bayesian calibration cambridge camera compass computation computer conf conference coughlan cvpr data dempster deutscher developments differentiation dimensional direction eccv editors extensions faugeras from geometric geometry griewank hartley iccv ieee image incomplete inference intl isard john journal june kluwer kosecka krishnan laird likelihood maccormick manhattan mathematical maximum mclachlan metrology multiple neural pages pattern press probability proceedings programming publishers recent recognition recovery references relative rotations royal rubin scenes series single society sons statistical statistics tanabe teller three university urban video view viewpoint vision wiley world yuille zhang zisserman
http://csdl.computer.org/comp/proceedings/cvpr/2004/2158/01/215810798abs.htm	106	An EM-like algorithm for color-histogram-based object tracking	algorithm anal analysis annals applications applied approach automatic azarbayejani ballard blob body bradski chapman chen collins color comaniciu commun component computer conf conference cvpr darrell data dempster density detection eccv editor estimation estimators face feature figure frame frames from fukunaga function gradient graphical hack hall hand heyden hinton holland hostetler huber human ieee image incomplete incremental indexing information intell intelligence interface international intl issue iterations iteratively jones jordan journal justifies kernel laird learning least likelihood lindeberg lncs location machine maronna maximum mean meer meth methodological minka model models monographs multivariate neal note number objects other outlier pages pattern pentland perceptual pets pfinder probability proc proceedings ramesh real recognition references regression results reweighted rigid robust royal rubin scale scatter selection sequence series shift silverman smoothing society some space sparse special squares statist statistical statistics summation swain techniques that theor theory through time toward tracking trans transactions tutorial understanding user using variants view vision wand welsch wiley with workshop wren
http://csdl.computer.org/comp/proceedings/cvpr/2004/2158/01/215810320abs.htm	43	Perceptual Organization of Radial Symmetries	accurate algorithm american another approach april atmospheric attention attentional attneave august based biomedical blobs boundaries cacm centers centroid circular computational computer conference context corresponding could coupled coupling curves cviu cvpr delineate delineation detailed detect detecting detection digital duda efficient elsevier engineeing essential evolution extraction face fast features figure framework free generalized grouping harmonic hart higherlevel hough ieee ijcv image images imaging immersion inferences information infrared interest january journal june later lead levine lines localization march mass means medioni memory minor more need normalization object operators original other pami parvin pattern patterns pictures points pose preprocessing proceedings process psychology radial real recognition references regularized reisfeld robotic robust scale segmentation sela september simply simulations sklansky society soille spaces step structures subceullar such symmetry tang that these time transaction transform validated verified vincent vision vortex voting watersheds which will with wolfson yang yeshurun zelinsky
http://csdl.computer.org/comp/proceedings/cvpr/2004/2158/01/215810002abs.htm	0	Using Plane + Parallax for Calibrating Dense Camera Arrays Vaibhav Vaish Bennett Wilburn Neel Joshi Marc Levoy	algorithms alignment anandan another aperture appearance applications approach array away based behind better beyond bmvc both buehler building bushes calibration camera cameras cart clearly cohen collins comparing complete computed computer conference constraints controlled correspondences criminisi critical cvpr cyclist depths different direct distances distributed does duality dyer dynamically eccv electronic eurographics everett experimental experiments facade factorization favaro field fields figure finite flexible focal focus focussed four frame frames from front full geomcalib geometry goal gortler graphics grzeszczuk hanrahan homographies horowitz http ibmr iccv ieee image images imaging irani isaksen kelin layout left length lens levoy light linux lncs lumigraph manor marvels matching mcmillan meters method methods metric microsoft mobile modelling motion moves moving mulitple multi multiimage multiple occluded occlusions original other parallax pattern people photography planar plane planes point possible practice proc proceedings produce produces projects ranging realtime recognition recovery reference references reid rendering reparametrized report research results right rigidity scene seeing seitz selfcalibration sequence sequences setup shape sharp sharpest showing siggraph smulski soatto society space spie springer standalone standing stanford stereo structure students sturm subspace sweep synthetic systems szeliski target technical technique tensors theory this through triggs true using vaish variable varied verlag video view vision volume weinshall well wilburn with workshop yang zelnik zhang zhengyou zisserman
http://csdl.computer.org/comp/proceedings/cvpr/2004/2158/01/215810135abs.htm	18	A Correlation-Based Model Prior for Stereo	active advances affinity algorithm algorithms anandan annals approxiieee april architectural associating august based belief berkeley biological birchfield boult boykov broadhurst brown burschka california cambridge camera carving cipolla complex computation computational computer conference contour correlation correspondence coughlan cuts cvpr data debevec december dense density depth detectable dissimilarity drummond eccv elements energy entropy estimation evaluation fast features frame framework friedman from function functions global gortler graph hager hastie hierarchical higgins iccv ieee ijcv image images inference information insensitive interpretation january july june kass kernel kolmogorov kumar layered learning longuet malik massachusetts matching mate mathematical mathematics matrix matsuura measure measurement measures minimization minimized mining mode models motion multiple multiway nakamura november occlusion occlusions ohta pages part parzen pattern patterns photographs pixel pointsampled polyhedral prediction press probabilistic probability problems proceedings propagation quadtree recognition reconstructing references registration regular regularization report sampling satoh sawhney scenes scharstein sciences scott september shade shum siggraph slanted snakes society space spline splines springer statistical statistics stereo sure surfaces symmetric symposium szeliski taxonomy taylor technical terzopoulos that tibshirani tomasi tpami tsin ullman university using veksler views vision visual volume what with witkin zabih zheng
http://csdl.computer.org/comp/proceedings/cvpr/2004/2158/01/215810090abs.htm	12	Multiview Occlusion Analysis for Tracking Densely Populated Objects Based on 2-D Visual Angles	academic activities aggarwal algorithm allocated although analysis angles appear appearances approach around association authors based bayesian believe blake burgard cadre camera carlo carving cluttered computer concept conclusion conference confirmed contribution cremers cvpr data davis densely description descriptions developed devised dockstader estimation exclusion experiments explicit feature field filter finding focuses formulation fortmann framework from fusion gaussian geometric gordon haritaoglu harwood have hull human humans hypothesized iccv icpr ieee ijcv image interacting ishiguro jain joint kutulakos laurentini limited localization maccormick major martin measurements methods mittal mobile monocular monte mori motion move moving mukawa multi multiple multiview needed nonlinear novel object objects observations occluded occlusion occlusions ocular ohya omnidirectional only open otsuka pami paper particle pattern people perez plane points populated presented press principle probabilistic proc proceedings process promising proposed real recognition recursive references region reid robot robustness salmond scene schulz segmenting seitz separately sequence sequential sethi shalom shape significant silhouette simultaneous situation smith society sogo space state stereo structure surveillance switching target targets tekalp that their theory this time tracker tracking trajectories trans understanding utsumi view views vision visual volumetric wang with workshop yachida
http://csdl.computer.org/comp/proceedings/cvpr/2004/2158/01/215810128abs.htm	17	Uncalibrated and Unsynchronized Human Motion Capture : A Stereo Factorization Approach P. Tresadern and I. Reid, Dept. of Engineering Science, University of Oxford, Oxford, England OX1 3PJ {pat,ian}@robots.ox.ac.uk	action active affine algorithm alignment analysis animation approach april articulated available barbara barcelona based blake body boston bregler british brooks buxton calibration california cambridge camera cameras capture carlsson caspi character chojnacki cipolla clusters complex computer conf conference constraints copenhagen correspondence correspondences costeira cvpr davis december defense degeneracies dependencies depth deutscher directed downloads dynamic editors eight engineering estimating estimation european exploiting exponential factorization factorizations feature foveated francisco free from full gavrila geometry goal hartley head hengel hilton http human humans ieee image implications intelligence international irani island journal july june kakadiaris kanade liebowitz machine madison malik manor maps markerless matching metaxas method metrology model modelling motion movement moving multi multiple murray norwich objects orthography overlapping pages pattern people point pooley press proc proceedings processing quan recognition reconstruction recovery references reid rigid santa scene scenes scheme self september sequence sequences sequenceto shape simakov single society spain step streams structure submitted synchronization synchronizing synchrony systems taylor techsheet their tomasi towards tracking transactions tresadern twists uncalibrated under university unsynchronized using vancouver vicon video videos view views vision volume voting wisconsin with wolf workshop zelnik zhou zisserman zomet
http://csdl.computer.org/comp/proceedings/cvpr/2004/2158/01/215810738abs.htm	98	Optimizing Motion Estimation with Linear Programming and Detail-Preserving Variational Method Hao Jiang, Ze-Nian Li, and Mark S. Drew School of Computing Science, Simon Fraser University, Vancouver, British Columbia, Canada V5A 1S6 {hjiangb,li,mark}@cs.sfu.ca	accurate adaptive affect algorithms algs analysis application approximate approximation barnard barron based basis beauchemin belief both boykov camera candidate chekuri chellappa chvatal comparison complexity computer computing conclusion conference correspondence costeira crucially cuts cvpr dense determining different dimension directions discrete disparity dissertation does eccv embedded energy equals estimate estimates estimation estimator ezra fast field figure fleet flow formulation freeman full functions general global graph graphs high horn iccv ieee ijcai ijcv image images inference ishikawa iterative journal kanade khanna kolmogorov labeling large left likelihood linear lower lucas maciel mapped matching maximum mechanism method methods metric minimization minimized model motion much multi naor ncamera occlusion occlusions olson optical optimization over pami paper parametric pattern peleg performance phase pose present problem problems proc proceedings programming propagation propose proposed pure ramin real recognition reconstruction reduction references registration relatively represent right rosenfeld same scale scaled scene scheme schunck search shown shum smaller society solution solve space spanned sparse stereo still stochastic structure symp technique techniques tensors texture than this time trans uniquely upper used using vectors veksler view vision visual vladimir werman what whole with without york zabih zheng zosin
http://csdl.computer.org/comp/proceedings/cvpr/2004/2158/01/215810826abs.htm	110	Multi-scale Visual Tracking by Sequential Belief Propagation Gang Hua, Ying Wu Department of Electrical & Computer Engineering Northwestern University 2145 Sheridan Road, Evanston, IL 60208	acknowledgment advanced analysis annealed approximate approximation articulated artificial automatic azur belhumeur belief blake body bottom brain cambridge camera capture carolina change comaniciu computer condensation conditional conf conference cote cvpr density deutscher dynamic dynamical efficient empirical european expectation face ference field fifteenth figure filtering foundation frame france freeman geometry grants graphical greece hager handbook head hierarchical hilton human ieee ihler illumination images indicated inference intelligence international isard island jaakkola jordan journal july june kerkyra large learning level localization loopy machine madison markov mathematical maximisation mean meer method methods mixed model modeling models motion murphy national network neural nice nonparametric north northwestern numbers objects octobor pampas parametric part particle pasztor pattern practice press primal probabilistic proc proceedings propagation ramesh real recognition references region reid result rigid sbpmc scale science september shape shift shum sketch sketchability society south state stockholm study sudderth supported sweden switching theory this time tobor towards tracker tracking trans tutorial uncertainty under using valued variational vision visual weiss willsky wisconsin with work zhang
http://csdl.computer.org/comp/proceedings/cvpr/2004/2158/01/215810098abs.htm	13	Difference Sphere: An Approach to Near Light Source Estimation Takeshi Takai Koichiro Niinuma Atsuto Maki Takashi Matsuyama Graduate School of Informatics, Kyoto University Yoshida-Honmachi, Sakyo-ku, Kyoto 606-8501 Japan {takesi-t,niinuma,maki,tm}@vision.kuee.kyoto-u.ac.jp	above acknowledgements acquiring adding ahuja albedo algorithm algorithms alleviated also although america application applications arbitrary artificial asano aspect availability based basri between bridging calibrating chellappa color complex computer conditioned conference configuration consider constraint contour contraction culture cvpr debevec detection difference dimensional direction distribution dynamic each eccv education effectively eklundh environment estimated estimating estimation even experiment extensions fifth figure filtering finding framework from generated generating geometry geotensity global goldgof grant graphics greenberg hanrahan have high horn hougen iccv identified ieee ikeuchi illuminant illuminants illuminating illumination image images imaging inaid integrated intelligence intensity into inverse investigate ipsj jacobs japan journal kambhamettu kimura known lambertian letters light lighting linear maki marschner ministry model multiple natural need nillius numerically object objects once onto optical other pami pass pattern pentland performs photography posed powell problem proc proceedings processing radiance ramamoorthi range rather real recognition reconstruction recovery reference references reflectance rendering representation research respectively samaras sarkar sato scene scenes science scientific shaded shading shape should shows siggraph signal signotes simple since single size society source sources sphere spheres sports stereo strategy subspaces superimpose supported surface surfaces synthesis synthesized synthetic technology that this three traditional trans under understanding varying virtual vision wang will wish with work yang yuille zhang zheng zhou
http://csdl.computer.org/comp/proceedings/cvpr/2004/2158/01/215810631abs.htm	84	A Variational Approach to Scene Reconstruction and Image Segmentation from Motion-Blur Cues Paolo Favaro Stefano Soatto Dept. of Computer Science, UCLA, Los Angeles, CA 90034, {favaro,soatto}@cs.ucla.edu	academic acha active adaptive aizawa algorithms alone animation anisotropic application applications approach approximately arbitrary artifacts assisted averaging based been bertero between blind blur blurred boccacci borman bottom boundary brightness bristol brostow burger camera captured center cetin chan changed chen choung circuits close compares comput computational computer conference contour contours contrast correspond corresponding curvature curve cvpr dame darker deblurring dependent depth different diffusion direct directions displacement dordrecht edges effect engl enhance essa estimate estimated estimation european evolution ezra favaro field figure filtering final fiume focus focussing foreground formulation from fronts generation georgeson gorea graphics gray hamilton hammett hanke heeger however hybrid icip ieee image images imaging initialized input institute intensities interval introduces introduction inverse jacobi jepson josa kang katsaggelos kaveh kluwer kopeika kubota lantzman left level lighter linear linearity local locations martin method methods midwest model most motion motions moving multiple nandhakumar nayar neubauer notre novel object objects obtained olsen original osher paik palm pami pattern peleg philadelphia phys physics points press problems proc proceedings processing propagating publishers publishing radiance recognition reconstruction recovered recovering references regions regularization regularized removal rendering research resolution restoration review right rigid robust scene segmentation segmented sequences sethian sharpening shifting shutter siggraph similar simultaneous smear snapshots soatto society spatially speed springs static stevenson stop subspace super surveillance symposium systems temporal this three times topright trans transactions tsai tull using variational vese view vision visualized volume wacv well were when while willsky with without workshop yitzhaky zomet zooming
http://csdl.computer.org/comp/proceedings/cvpr/2004/2158/01/215810804abs.htm	107	Covariance-Driven Mosaic Formation from Sparsely-Overlapping Image Sets	afterwards aids airlie algorithm alignment allied anal appearance application applied assisted assoc band based beaton before bootstrap bottom brown cambridge canada care center change chapter classification closest computer computing conf conference constraint construction correction covariance curved cvpr data degrezia diabetic disease disorders distortion driven dual dualbootstrap early eccv edition estimates estimation evaluation evans experiment feature features field figure fitting flannery from generation geometry global gregory group hartley herman hierarchical high hirvonen house human iccv ieee ijcv illustrated illustrating image images imaging inference intervention inves irwin iterative joint june kumar left lens linear local lowe majerovics manifestations manifold matei meaning miller misalignment models modification montreal mosaic mosaicing mosaics mosby multi multiple muse neter number numerical nurses ophth ophthalmic ophthalmology order oriented pages pairs panoramas panoramic paper pattern peleg pigmentosa point polynomials power presence press proc proceedings projection recipes recognising recognition reference references regions registering registration report research results retina retinal retinitis retinopathy richard ries right robinson robust roysam sawhney scale scientific sections show shown shum society sparse spectroscopic statistical stewart study subset surface systems szeliski tanenbaum technique technometrics teukolsky their through topology trans transformations treatment true tsai tukey unbiased university update using vetterling video view vision volume wasserman weleber wildes with zisserman
http://csdl.computer.org/comp/proceedings/cvpr/2004/2158/01/215810386abs.htm	52	Shape Constrained Image Segmentation by Parametric Distributional Clustering	able accuracy achieved active address adversary aggregation algorithm almost also amount analysis animal apply approach area aspect aubert background band barlaud base based basri bayes bearing because been besson between boston boyle brandt buhmann case cheetah cheetahs closely clustering color column combining comparable comparison competition comprises computer conclusion conditions conference considerable constrained constraints containing contours contributed correct covered cremers cvpr dark data dempster depicts differences discriminative distributional does eccv edition editors elements estimation example fact faugeras features figure filter filters foreground foreleg form fragmented framework from front galun gradients green group growing have head hermes heyden higher histogram hlavac iccv identical identified ieee image impression incomplete incorporate individual information intelligence irregularities jehan johansen journal knowledge kohlberger laird lead lecture level likelihood lncs machine mass match maximum method mind missed models moreover most motion multi multiscale mumford nevertheless nielsen nonlinear notes only pages paper paragios parametric part parts pattern plain pose prior priors probability problem proceedings processing proposed recognition references region regularization remarkable representations response responses restrictions results rousson royal rubin schnorr science second segment segmentation segmentations semantically shadows shah shape sharon silhouette small snakes society some sonka sparr springer statistical statistics tail terms texture that these this transactions understanding unifying using variational very vision volume while with without yuille zoller
http://csdl.computer.org/comp/proceedings/cvpr/2004/2158/01/215810356abs.htm	48	Estimation of Blood Flow Speed and Vessel Location from Thermal Video	acad accuracy achieved additional adjoint aimed alternatives analysis anatomical anatomy annals annual applied arkin arterial arteries associated average ayaswamy baish bardon behavior bioheat biologic biology biomech biomechanical biomedical bleed blood body branching brinck buse cancun capitalizing channel chieti circ close compagnon comparison compute computer conceptualization conclusion conduction conduit conference conservation consideration continuous control convergence cooper costentin countercurrent cvpr december deception dependent describe detail developments difference dilatation dimension dimensional dimensionnal dimensions donato dynamic eberhardt effect ekpene electron emotional energy engineering engr equation estimation experiment face fear flow forearm foster foundation functional fundamental gender general geometry gibbs gottlieb have heart heat holmes human iacob ieee image imaging increased influence infrared inhomogeneity international into invasive inverse italy iwatani january jarny jiji joannides journal june keller lagarias lahary lancet large lemons levine living local magazine mass mead meaning means measurements mechanisms medical medicine merla method mexico microstructure model modeling monitoring more multidimensional muscle nature nelder network november optimization ozisik paper parameters part pattern pavlidis pennes perfused perfusion peripheral phenomenon physical physiol physiological physiology pixel polygraph preliminary proceedings profiles properties rate reactions recent recognition recording reeds references resolution respect response responsible resting results review revisited role romani rossini scoring seeing seiler september serve should siam simplex simplified skin society solving source study surface sympathetic take temperature temperatures term testing theory thermal this three through thuillez tissue tissues trans transactions transfer transport upon used using values various vascular vessels video vision wang weibaum weinbaum werner wissler with wright wulff york
http://csdl.computer.org/comp/proceedings/cvpr/2004/2158/01/215810520abs.htm	69	Efficient Tracking with the Bounded Hough Transform	abraham accommodate acknowledge acknowledgements acquisition algorithms allows also approach arbitrary aspect atiquzzaman attractive authors automation autonomous ballard been being benefit benefits besl between blais both bounds california canada capture carter cgvip chengping coherency compare comprising computational computer conclusions conf conference correctness create cube cvpr damper daniele data david degree demonstrated dense detect detecting determination diego distance dixon domenico dynamic effectively efficiency efficient error estimation executes experimental exploit extended extraction fast feature february figure filter filtering financial formulation frame frames francois free freeform from functions future general generalizing geometric godin gratefully greenspan grimson have hebert hierarchical high hough huttenlocher hybrid idea ieee illingworth images implement implementations including increase increased inter international intl investigated iocchi isairas jasiobedzki john june kalman kanade kittler known lappas localization ltrs luca mahalanobis main march martial mastrantuono mckay mdrobotics measurement method michael michel model montreal motion moving multiresolution nardi nash neil novel nserc object objects only optimization orbit outliers pages pami particle pattern patterns paul pelopidas performance performed picard plan points pose possible possibly precision predictive presented probabilistic probing proc proceedings prove quebec range rate real realtime recognition recursive reduced references refinement registration relationship results robert robotics robustness roth satellite sensitivity sensor sequence sergio shape shapes shapres simon simulated size small society space sparse speed starts such support surfaces survey switches symp symposium takeo talbot technique techniques tens tests that then track tracking trades trajectory trans transform transformation using velastin velocity verify verion very vision well will wish with work works
http://csdl.computer.org/comp/proceedings/cvpr/2004/2158/01/215810769abs.htm	102	The Multibody Trifocal Tensor: Motion Segmentation from 3 Perspective Views	algebraic analysis anandan applications approach automated bayesian body bolles cambridge cartography communic component computer conditions conference consensus costeira cvpr data eccv entities extraction factoriza feng fischler fitting from generalized geometric geometry gpca hartley homography huang iccv ieee ijcv image independently integrated kanade kanatani kosecka layer london matrix method missing model motion moving multibody multiple objects pages pami paradigm pattern perona perspective phil planar points powerfactorization principal proceedings random rank ransac recognition references represent royal sample sastry scene segmentation selection separation sequences shashua soatto society static structure subspace szeliski tensors that tion torr trans unified using vidal view views vision volume with wolf zisserman
http://csdl.computer.org/comp/proceedings/cvpr/2004/2158/01/215810444abs.htm	60	The World in an Eye	ability about accurate acknowledgement adler affect also analysis application applications baker barsky based behavior believe body bolt burkhard calibration camera cameras capability captured capturing catadioptric caustics center central chapter circumstance clinical color columbia company computed computer conducted conference confidence consciousness context conveys cornbleet cornea corneal corners corresponding curved curves cvpr dang data daugman density department dept depth details determine direct directions distribution easily edition editors ekman eleventh emotions environment epipolar errors estimating estimation expressions extend eyes face facial fields figure fitting flom flux following formation four frame framework frey from frontal funded gation gaze gemoetrical geometry germany give grant graphics great grossberg halstead hamburg helmholtz high hlava human hutchinson iccv identification ieee ijcv illuminating illumination image imagery images imaging implications important independence infer information input insights instance instrumentation intent interaction interactions interface interfaces internal into invasive iris jacob john josa just kaufman klein left light lights location london look looking machine makino mandell medical methods microwave mirrors miyake more mosby movement nayar nene networks nishino normals objects only optical optics oxford pages pajdla panoramic passive patent pattern patterns perfect person physiologic physiology pointer possible powerful preparation press proc proceedings propoa reaction recognition reconstruct reconstructed reconstructing record recording recovered recovering references reflection reichert relight report research results reveals right robots rosenberg rotation royal safir scene science sensing sensors serve shealy sheena side siggraph single social society sons sources specular sphereo spheres spherical spie spline springer statistical stein stereo structure studies such sung surface surfaces surrounding surveillance survey svoboda swaminathan synchronization system teaching technical techniques tells test than that theory third this through tomkins tool tpami tracing tsmc tsumura university useful using valuable value venkateswarlu verlag very view viewpoint vision visual visualizes volume voss wang well westheimer what when white wiley wire with work york young
http://csdl.computer.org/comp/proceedings/cvpr/2004/2158/01/215810625abs.htm	83	Video stabilization as a variational problem and numerical solution with the Viterbi method	achieve acoustics adaptively additional algorithm algorithmic algorithms allowed although amini amount analysis applications approach arbitrary august automatic band based been between bosse both buehler build camera causality change characterization chellappa clues compensatory complex computation computer conference constrained constraints consumer cope could cvpr data december deeper define depth different digital discussed discussion dynamic electronics error estimates estimation evaluation experimentally experiments explored extensions extrapolation features filter filtered forcing formal formulated forney foundations frame frequent full functions global gone have hidden houston ieee image inherent instance intelligence intentional interesting international interpolation into investigated involve ishii jain kitamura labs lavalle length long machine march markov mcmillan method metric minimal missing models more morimoto morimura motion narrowing navigation nonetheless notable numerical obviously only optimal optimality order others output over pages paper parameter particular partly pass pattern performance perhaps pilu possibility possible potential predicting presented preserve previous problem problems proc proceedings processing producing programming properties quasi rabiner real recognition references related relationship rendering report response robotics robust sakaguchi sample samples satisfactory section selected sept sequences setting shown signal simplicial since situations society solution solving speech stabilization stabilized stabilizing stage states still study subtle system technical thanks that their thereby this through time tolerance total trajectory transactions trellis tuning tutorial unsuccessful uomori using value variational video vision viterbi volume wearable weymouth whilst widening with within workshop would
http://csdl.computer.org/comp/proceedings/cvpr/2004/2158/01/215810143abs.htm	19	Shape Correspondence through Landmark Sliding Song Wang, Toshiro Kubota, and Theodor Richardson Department of Computer Science and Engineering University of South Carolina, Columbia, SC 29208, U.S.A. songwang@cse.sc.edu, kubota@cse.sc.edu, richa268@cse.sc.edu	able according accurate achieve active after algorithm aligns along also anal analysis application applications applying approach april automatic barden based before belongie between bookstein both carne chui classification closed compact comparisons components comput computational computer conclusion conf conference considered consistency construction contexts contours cooper cootes correspond corresponded correspondence cost crossing curve curves cvpr davies decomposition defining deformation deformations deformed description developed differences different dimensions diplaros direct discuss distorted dynamic eigenshape except faugeras features figure first flexible forms fourth framework from function gdalyahu generate geodesic geodesics geometric geometry global graham grimson group hierarchical hill identification identifying ieee imag image influence initializa instances intell into issue john june kendall kotcheff landmark landmarks left length leventon like local machine malik mapping march matching medical method methods milios minimum modeling modelling models modles more morphometrics multi multiple nonrigid novel object occluded open optimal optimization outline pages paper pattern peterson petrakis plate point points powell practical presented previous principal proc proceedings programming proposed puzicha rangarajan recogn recognition references respectively results retrieval right rigid second self shape shapes show silhouettes similar sliding small society sons spline splines springerverlag staib statistical surface syntactic taylor techniques that their theory thin third this those three training trans twining underlying understanding used using varied variety varying vision wang warps waterton weinshall wiley without
http://csdl.computer.org/comp/proceedings/cvpr/2004/2158/01/215810220abs.htm	29	Fast Contour Matching Using Approximate Earth Mover's Distance Kristen Grauman and Trevor Darrell Computer Science and Artificial Intelligence Laboratory Massachusetts Institute of Technology Cambridge, MA, 02139	abbasi access algorithm alon approach approximate assignment athitsos based bases belongie bmcv boostmap canada carlsson categories change character cohen computational computer conf conference content contexts contour copenhagen corfu correspondence curious curvature cvpr data demirci dickinson dimensional dimensions distance dvir earth eccv edinburgh editing efficient egisys embedding embeddings enable fast flow france geometry giannopoulos gionis gold graduated graph graphs gray greece greenspan grouping guibas hagedoorn hashing high iccv ieee ijcv image indexing indyk intl keselman kimia kittler klein kollios labs large level libraries lihue madison malik many matching method metric mokhtarian mori motwani mover nice object order pattern peleg point poser proc proceedings pseudo puzicha rangarajan rankings recognition references region report resolution retrieval robust rubner scale sclaroff search sebastian sept sets shape shapes shock shokoufandeh sicily similar similarity society solution space stanford state statistical structure tech thaper theories thesis through tomasi tpami transformation ultimate under unified university using utrecht veltkamp very video vision washington weighted werman wkshp
http://csdl.computer.org/comp/proceedings/cvpr/2004/2158/01/215810594abs.htm	79	Scale Selection for Anisotropic Scale-Space: Application to Volumetric Tumor Characterization	algorithm anal analysis anisotropic artificial automatic bandwidth based biol cancer cartesian characterizing comaniciu comput computer conf conference consider covariance cvpr cybern dalal data deriche detection differential diffusion driven edge errors feature features filtering filters florack frequency griffin haar have henschke heteroscedastic ieee image images imaging intell invariants joint kanatani kanazawa karlsruhe koenderink krishnan lillholm lindeberg local lung machine malik matei math mathematical matrices mccauley mcguinness miettinen mirtcheva models nielsen nodules nonsolid okada pages part pattern perona prague proceedings really recognition references regularization robust roentgenol romey rutgers scale screening selection significance society solid space structure structures thesis trans university using vancouver variables viergever vision with witkin yankelevitz
http://csdl.computer.org/comp/proceedings/cvpr/2004/2158/01/215810470abs.htm	63	Grouping Dominant Orientations for Ill-Structured Road Following Christopher Rasmussen Dept. Computer & Information Sciences University of Delaware Newark, DE 19716 Abstract	acoustics adapting adaptive adelson advanced agency analysis apostoloff applications approach architecture asian asilomar automated automatic automation autonomous available based blake brady cascaded challenge classication color combining comp components computer computers computing conditional conf conference contour crisman crossing cues curvaturebased cvpr darpa defense density design detection efficient estimation european expert feng fftw filtering filters finding following freeman frigo from gabor gool grand grandchallenge grouping guidance hancock harvester hough http ieee image images intelligence intelligent intl isard issue jain jochem johnson lane laser lines local machine malik masaki milanfar multiple multiscale nagel ollis orientation parallel partially particle pattern perception perspective planar point points polk pomerleau pose principal proc proceedings processing proesmans projects propagation range rapidly rasmussen real recognition references representation research ribeiro road roads robotics robots robust scene scenes schaffalitzky segmentation shape sighted signal signals skew society software southall special spectral speech springer staircases steerable steering stentz stereopsis stochastic symposium system systems taylor texture their thorpe time tracking trans transform turina tuytelaars under unscarf unstructured urban using vanishing vehicle vehicles vision voting wavelets weber zebra zelinsky zhang zisserman
http://csdl.computer.org/comp/proceedings/cvpr/2004/2158/01/215810283abs.htm	38	Joint feature-basis subset selection Shai Avidan Mitsubishi Electric Research Labs 201 Broadway Cambridge, MA 02139 avidan@merl.com	advances algorithm analysis application applied artificial beyond bins blake blum boosted bottou bound branch brooks burges cambridge cascade case categorization chapelle classifier cole comparative comparison computationally computer conf conference copenhagen cortes cvpr data decision denker denmark detection digit discovery draper drucker editor effectiveness efficient eleventh european examples face feature features floating fourteenth freund from fukunaga girosi green grove guyon handwritten hawaii huge icml ieee information intelligence international irrelevant jackel jerusalem joachims john jones journal kaufmann kernel kernels kittler knowledge kohavi langley large learning lecun letters machine machines making marill mateo methods mining morgan mukherjee muller multi munich narendra nature networks neural nips novovicov object oldenbourg optimization osuna pacific pages pattern pedersen pfleger poggio practical press problem proc proceedings processing pudil puerto rapid rawlings receptors recognition references regression regularization relevant rico romdhani rules sackinger saitta scale schlkopf scholkopf search selection sets shum simard simple simplified smola society springer statistical study subset support svms systems text theory torr training transactions tutorial using vancouver vapnik vector verlag view viola vision volume wadsworth weston with wrappers yang zhang
http://csdl.computer.org/comp/proceedings/cvpr/2004/2158/01/215810010abs.htm	1	Restoration of Curved Document Images through 3D Shape Modeling Zheng Zhang, Chew Lim Tan, Liying Fan School of Computing, National University of Singapore 3 Science Drive 2, Singapore 117543 Email: {zhangz, tancl,fanly}@comp.nus.edu.sg	accuracy acknowledgement adjust agency algorithm alignment also analysis applicable arbitrarily area assumption august average axis baird based beyond book bound brown comparison components computer conclusion conference contrast copying correcting could curl curved cvpr cylindrical deblurring december defect dengle describe deskewing ding discuss distorted distortion distortionfree document documents education efficient evaluation experiments explored figure from further future general grant have hoch iapr ieee image images implement improved improvement improves international interreflections into journal july june junker light lines look machine marked matsuyama ministry minute model models normal objectives obtained october order original outputs page paper parallel pattern pilu precision present problem problems proceedings process processing providing proximal real recall recognition recovering rectify references regression relaxing remove research resolution respectively restoration restore restored results runtime same scanned scanner scanning science seales seconds shade shading shape show singapore society solution solves source speed spine structural successfully supported surface surfaces syntactic system table technology test text that their these thick this thus time tolerable ukida under undoing unfolded uses using vision volume wada warped warping which while will with work works workshop world zhang
http://csdl.computer.org/comp/proceedings/cvpr/2004/2158/01/215810058abs.htm	8	Autocalibration & 3D Reconstruction with Non-central Catadioptric Cameras	accurate accv adjustment algebraic aloimonos ambiguous angle applications applied applying approximate approximating assumed assuming auto automatically axis bajcsy baker based baseline been benosman bird bmvc bonfort both bottom brodsky bundle calibration cambridge camera cameras carving catadioptric caustics center central chum compute computer conference coordinate correspondences corresponding cvpr daniilidis demmel derrien described describes design detail detected detection devices digital directions distortion dongarra easily editors eigenvalue electronic epipolar essential estimated estimation ever exact explanation extremal fermuller fields figure film finding first fitzgibbon fixed focal formation france from gachter general geometry germany geyer grossberg guide hardly hartley have hicks hyperbolic iccv ieee ijcv image imager images imaging improved initial inliers input internal intersection kang konolige leads left length lens lenses light linear manually many marked matas matrices matrix maximally method metric micusik middle mirror mirrors model models more motion multiple nayar neumann nice noncentral obtained oliensis omnidir omnidirectional optical optics orientation outliers oxford pages pair pajdla pami panoramic para parabolic parameters pattern perspective philadelphia photographic photography pixel plane plenoptic pless point position positions practical press problems proc proceedings projections quality radial rays recognition reconstruction reconstructions rectangular references regions rejected relative resolution results right robust room ruhe same second sensors shape show siam simultaneous single skew small society solution some space specular spherical springer square stable stereo structure sturm surfaces svoboda swaminathan symmetry system systems templates that theory these together tuning uncalibrated university urban used using variant verified verlag very video view viewpoint views vision vorst voxel were what wide with workshop zero zisserman
http://csdl.computer.org/comp/proceedings/cvpr/2004/2158/01/215810834abs.htm	111	Collaborative Tracking of Multiple Targets Ting Yu and Ying Wu Department of Electrical & Computer Engineering Northwestern University 2145 Sheridan Road, Evanston, IL 60208 {tingyu,yingwu}@ece.northwestern.edu	academic active aerospace among april around association behind blake blue cadre carmichael computer conf conference contours cvpr data davis details detecting drilling electronic face figure filtering fortmann freeman gesture haritaoglu harwood ieee illustrate isard japan journal learning links london lowlevel markov mfmc moving multiple nara network objects orlando particle pasztor pattern people perez players please press proc proceedings real recognition references shalom soccer society springer static structure system systems targets tennis three threehumans time tracker tracking transactions twohumans verlag vision walking what when where with women
http://csdl.computer.org/comp/proceedings/cvpr/2004/2158/01/215810819abs.htm	109	Reconstructing 3D Independent Motions Using Non-Accidentalness K. E. Ozdena K. Cornelisa L. Van Goolab	accidental accidentalness acknowledge acknowledgement action again alternative ambiguity among analysis applicability applications approach arbitrate artificial assumption attention authors automatic avidan based basis battery between biermann body both brand bregler cambridge camera case challenge combination community components computer conclusion conference consistency constraint constraints construction contain containing copenhagen cornelis correct costeira council criterion cuts cviu cvpr demonstrated different direction dynamic eccv effect estimation even expect experiments explore eycken fact factorization finally fitzgibbon flemish focus forms formulation found from fund fundamental future gave generally generic geometric geometry gives gool gratefully hartley heading hertzmann iccv ieee ijcv image images independence independently intelligence interactive investigated irani issue july kanade leuven life like likely line london lost lowe machline makes malik manor many marvel mathematically matrices matrix mentioning method might model modeling models monocular more morphable motion move moving much multi multibody multiple need nonaccidentalness nonrigid normalized november object objects observation orthography ozden pages pami paper parallel parts pattern periodicity perspective phil planar planarity plane planes point points possible press principle problem proceedings project projection properties property proposed real receive recognition reconstructing reconstruction reconstructions recovering references relative remains representation research results revisiting rise royal sastry scale scales scene scenes scientific search second segmentation selection sequence several shape shashua simple simultaneously since single soatto society solutions some special states streams structure sturm such support system techniques that their them therefore this those tomasi torr tracking trajectories trajectory trans translation travels triangulation uncalibrated under university using vancouver very vidal video view views vision visualization when which will with wolf work workshop worthwhile would wrong zelnik zisserman
http://csdl.computer.org/comp/proceedings/cvpr/2004/2158/01/215810394abs.htm	53	Globally Optimal Segmentation of Interacting Surfaces with Geometric Constraints	accurate acknowledgment adaptive airway airways algorithm algorithms american anal analysis analyzed anatomical application applications approach august authors automata barrier been beyond border boykov boyle branch brooks cambridge cardiac cardiology center chen cineangiograms city closure cole collins colloquium combinatorial company comparison computational computer computers computing conclusion conference cross cvpr data decomposition deformable desired detecting detection detector developed diameter dimensional dimensions dissertation dynamic edinburg efficient emmcvpr employed energy environment eric evolving experimental extensible fast fedkiw fleagle flexible flow fluid frank from functions generated geoffrey geometric geometry globally goldberg grant grants graph graphs guidance higher hlavac hochbaum hoffman human icalp ieee imag image imaged images imaging implementation implicit information inherent inner intell interfaces international interrelations intravascular iowa jacm journal juerg july june kolmogorov labeling languages level machine magnetic malaga management marching master matching materials maximal maximumflow mclennan means mechanics medical method methods mincut minimization minimum modeling models multi multiple mutually networks nhlbi objective optimal osher outer pages parameters part pattern performed phantom physical picard point polynomial preserving press prince problems proc proceedings processing programming properties provide providing publishing pulmonary quantitative readily recognition reddy references related research resonance respect resulting robust sagittal science search searching sections segment segmentation segmented separation sept sequences sequential sethian simultaneously skorton small smoothness society sonka spain spie springer springerverlag supported surface surfaces symposium technique technology testing texas thank thedens theory thesis third three time topology trans transversal tschirren ultrasound under university using validated validation various verlag vessels view vision volumetric walls winniford with workshop
http://csdl.computer.org/comp/proceedings/cvpr/2004/2158/01/215810066abs.htm	9	Window-based, discontinuity preserving stereo	acknowledged adaptive agrawal algorithm algorithms alibhai analysis application applying approximate based baseline bayesian belief binary binocular birchfield bobick boykov chai compact computation computer computing conference contour contract corner correspondence curve cuts cvpr cycle data davis december dense different diffusion dimensional discontinuities dissimilarity distributions duraiswami dynamics efficient electron embedded energy epipolar estimation european evaluation evaluations exact experiment fast fifth figure financial flow formulation frame framework from functions fusiello geiger geman general gibbs global golland graph graphics graphs gratefully greig ground guided handling harwood ieee image images indian insensitive intelligence international intille ishikawa journal july june kanade kang kolmogorov kumar ladendorf large level lines lncs luther machine matching matting maximum measure microscope minimization minimized minimum modeling motion multi multiple multiway ncamera nonlinear november occlusion occlusions okutomi optimization pages pattern pixel pollefeys porteous posteriori problem proc proceedings processing progressive propagation putting ratio recognition reconstruction references relaxation restoration result results roberto robotics royal sampling sawhney sawtooth scene scharstein scheme seheult sept september series sets seventh shan shum sixth size slanted smile society springer statistical stereo stochastic structure support surfaces systems szeliski taxonomy that theory thesis three tilt together tomasi transactions transmission transparency trucco truth tsukuba ultrastructure uncertainty under university using vekseler veksler venus verlag view vision visual volume website what window windowing windows with workshop york yuille zabih zhang zheng zucker
http://csdl.computer.org/comp/proceedings/cvpr/2004/2158/01/215810030abs.htm	4	Algebraic solution for the visual hull	advances algebraic approach based baumgart beijing bmvc bombay boundary bounded boyer brand british buehler cambridge canada carving charting china cipolla complex computer computing concept conference contour contours cooper cross curves cvpr dual duals eurographics european exact field fishman fitting franco from geometric geometry giblin hull hulls hybrid ieee image india information international invariants january journal kang kriegman kutulakos laurentini lazebnik light linear machine manifold matusik mcmillan modeling motion neural nips object objects occluding pami pattern polyhedral ponce press proc proceedings processing quadric real recognition reconstruction references renaudie rendering seitz shape silhouette singular smooth society solids space stanford surface surfaces systems tarel their theory thesis time understanding unified university using vancouver vision visual workshop zisserman
http://csdl.computer.org/comp/proceedings/cvpr/2004/2158/01/215810050abs.htm	7	Spherical Harmonics vs. Haar Wavelets: Basis for Recovering Illumination from Cast Shadows Takahiro OKABE, Imari SATO, and Yoichi SATO Institute of Industrial Science, The University of Tokyo 4-6-1 Komaba, Meguro-ku, Tokyo 153-8505, Japan {takahiro,imarik,ysato}@iis.u-tokyo.ac.jp	accv acknowledgements acquiring advances alignment alleviates analysis another appearance appendix applied approach approximation arbitrary assuming band bandwidth based basis basri because belhumeur between bidirectional bouguet bouguetj brdf bridging calibration caltech cambridge camera cast change choice cient coefficients cohen color comparisons components computer computing conditions cone conference convex convolutions course culture cvim cvpr debevec defined degree degrees dependent derive derose determining diffuse direction directions distibutions distribution driscoll dynamic easily eccv education efficiently environments equation estimation even expansions experimental face fifth finally flannery fourier framework frequency from function functions future generally geometric geometry georghiades given global gortler grants graphics greenberg haar hanrahan harmonics have healy hence hierarchical high hoffman iccv ieee ikeuchi illuminants illumination image images imaging important integral integrand integration interactive into inverse ipsj irradiance issues jacobs japan japanese kaufmann kautz known kriegman lambertian legendre less lighting limitations limited linear lncs machines making many maps marschner mathematics matlab method methods might miller ministry modeling models more morgan most multiple namely nishino note notes numerical object objects obtaining okabe onto optimal orthogonal pami parameters part pattern photography photometric play polynomial pose potential precomputed press proc proceedings processing product proposed provides publishers radiance ramamoorthi range ransac real recipes recognition reconstruct recovering references reflectance reflection regard relationship rendering represented representing research respect respectively role salesin samaras sampled sampling sato scene scenes schroder science scientific shadows shashua shum siggraph signal similar simulated single sloan snyder society sophisticated sparse specular sphere spherical sports stable stollnitz subspaces substituting such superimpose support supported sweldens symposium synthesis synthetic technology teukolsky textured than that therefore this those thus time toolbox traditional trans transfer transform transforms under uniquely university used using variable variational varying vector vetterling view virtual visibility vision visual wang wavelet wavelets weight where wipl with work zhang
http://csdl.computer.org/comp/proceedings/cvpr/2004/2158/01/215810240abs.htm	32	Geometric and Shading Correction for Images of Printed Materials A Unified Approach Using Boundary Yau-Chat Tsoi and Michael S. Brown Department of Computer Science H.K.U.S.T. Clear Water Bay, Hong Kong {desmond, brown}@cs.ust.hk	academic addition additional adelson adopted advances aided algorithm allowing allows analsysis angle applicable apply approach archival artifacts automated available avanidra barrow based between binary binder blending bound boundary brockington brown burden camera captured care chamzas changing characterisitcs chaudhuri color common component composed computer conclusion conference context coons correct corrected correction could curl currently curve curves cvpr cylindrical deal degradation depth deriving design detection determination diego digitization digitized ding distorted distortion distortions document documents drew each easily eccv efforts environment especially estimate estimation experience facilitate farin first folding folds found four framework freeman from funt gatos general generate geometric global great greece guide haralick have iccv icdar icip ieee illumination image imaged images imaging improved including incorporated information innovations input international interpolation interreflections into intrinic intrinsic item japan jiang july kanungo length letters light line local mapping material materials matsuyama model modeled models national near need neural nips novel obtained october only opposite orientation page pages papamarkos paper parameterization parameterize parameterized patch pattern personal phillips physical pilu position present presented presents preservation press printed proc proceedings process processing projection props provides proximal purposes reasonable reasonably recognition recovering rectify rectifying references remove removed report restoration restored restortation robust routinely ruled rulers same scanned scene seales secondly sequences service several shading shape show showed significant simultaneously single skew society source spent springfield strips such summary surface surfaces systems tappen technical technique techniques tenenbaum text that thessaloniki thick this time transactions tsukuba ukida under undoing uniform used using variety vision wada weiss which with without working would zhang
http://csdl.computer.org/comp/proceedings/cvpr/2004/2158/01/215810722abs.htm	96	A Unified Spatio-Temporal Articulated Model for Tracking Xiangyang Lan Daniel P. Huttenlocher {xylan,dph}@cs.cornell.edu Cornell University Ithaca NY 14853	academic analysis appear applications articulated automatic based black bobick body bottom bregler cardboard cipolla collins computer conf conference cvpr davis dynamics elgammal elschlager estimator exemplar face felzenszwalb figure filtering finding fischler forsyth france from gait gavrila gesture gross hidden human huttenlocher iccv identification ieee ijcv image international january juan june learning linear maccormick madison markov matching model models motion movement multi nice nips object october pages pami parameterized parametric pattern pavlovic people pictorial press proceedings puerto rabiner ramanan recognition recognizing references rehg representation results rico selected sequence sequences shape shet silhouette society speech stenger structures survey switching thayananthan torr tracking trans tree tutorial understanding using video viewpoint vision visual wilson wisconsin yacoob
http://csdl.computer.org/comp/proceedings/cvpr/2004/2158/01/215810611abs.htm	81	Visual Odometry and Map Correlation Anat Levin and Richard Szeliski Microsoft Research	absolute additional adjustment algorithm algorithms allen also alternation alvey appearance applications approach approximate april arbitrary architectural artificial atlas average based beis belief blaer bosse building bundle cambridge camera cameras capturing cipolla combined comp computation computer conference constraints context corner correctness correlation correspondence coughlan could cvpr deformable detector development dimensional distance drawn eccv edge edition efficient entail environments essa estimation exploration exploring fact fast ferreira final finding flat floors followed framework freeman from future gaussian geometry golub graphical graphics hand harris hartley high histograms holiday hopkins iccv icra identify ieee image immersive improved indexing indoors inertial information innovative intelligence interactive intl investigate john june kaufmann large least like linear loan localization loopy lowe many mapping matching matrix measurements method metric millenium mobile model models modern more morgan motion multi multidimensional multiple nearest neighbour neural nourbakhsh object odometry optimal optimize organize orientation panoramic pattern peleg place plan pless points press problem problems proceedings processing propagation provided quality real recognition recomputing references relies results robertson robot robotic rosenfeld rough scalable schaffalitzky search sensors sept sets shape shapes similar singh snaps society solving spaces squares stage steedly stephens strelow structure such survey synthesis system taylor techniques temporal that third this thrun tilted topological topology torralba trajectory trans triggs ulrich university unordered using uyttendaele vanishing very videoplus view views vision visual visually wacv weiss werman where which will with work workshop world would zisserman
http://csdl.computer.org/comp/proceedings/cvpr/2004/2158/01/215810604abs.htm	80	Error Analysis for a Navigation Algorithm based on Optical-Flow and a Digital Terrain Map	accuracy advances aerial ahuja algorithms analysis azriel bowyer computer conference correspondence correspondences covariance critique cvpr david dementhon determination digital duraiswami eccv elevation estimation feature from haralick higher huang ieee image integrated intelligence international interpolated john journal lncs machine models motion netravali oliensis park pattern pose position press proc proceedings propagating recognition rees references remote resolutions review rosenfeld samet sensing sequences simultaneous society softposit structure transactions understanding using vision washington
http://csdl.computer.org/comp/proceedings/cvpr/2004/2158/01/215810195abs.htm	26	Camera Network Calibration from Dynamic Silhouettes Sudipta N. Sinha Marc Pollefeys Leonard McMillan. Dept. of Computer Science, University of North Carolina at Chapel Hill. {ssinha, marc, mcmillan}@cs.unc.edu Abstract	across adjustment ahuja algorithm algorithms alignment application approach astrom august baker based baumgart bmvc bolles bounded boyer buehler bundle calibration cambridge camera capture carving cheung chum cipolla circular combining computer computing concept conf conference constraints construction continuous convex corfu curve curved cvpr cylinders data deformation determining dissertation dominant dynamic eccv epipolar estimation eurographics european exact features finding fischler fitting fiztgibbon flexible from frontier generalised geometric geometry giblin gool gortler graph greece hartley hebert heyden hull hulls iccv ieee ijcai ijcv image information johansen joshi july june kanade kriegman kutulakos laurentini lazebnik lecture letters levi line lncs matas matching matusik mclauchlan mcmillan melkman mendonca model modeling modern monocular motion multiple nielsen notes objects orientations pages pami pattern pencils perspective plane planes pollard pollefeys polygon polyhedral ponce popovic porrill practice press proc proceedings profiles projection randomized range ransac raskar real recognition reconstruction recovery references refinement rendering review sand scenes schmid science seitz sethi shape shapes siggraph silhouette silhouettes simple skin smooth soatto society solids space sparr springer stanford stereo structure surfaces surviving synthesis szeliski tangent test theory time trans triggs uncalibrated uncertainty under understanding university unknown verbiest verlag view viewing vijayakumar vision visual werman with without wong workshop yezzi zhang zisserman
http://csdl.computer.org/comp/proceedings/cvpr/2004/2158/01/215810334abs.htm	45	3D Models Coding and Morphing for Efficient Video Compression	adjustment amsterdam analysis application applications applied approach architecture automatic based between california camera capture chavira closed coder coding comparison compressed compression computer conference crack cvpr debevec deguchi deknuydt dimensional dublin eccv electronic emerging estimation eurasip fdis figure fitzgibbon from galpin geometryand girod gool graphics head hierarchy hybrid icip ieee image imagery images imaging information ireland japan joint jose journal jpeg june koch kochand korea left lines magnor malciu malik marquant martinez meshes method model modeling models morin mosaicking multiviewpoint nara natural nister object open orleans pateux pattern performance photographs pollefeys pose position preteux proc proceedings processing psnr rate realistic recognition reconstructed reconstruction recovery references rendering representation right scene schwarz seoul sequence sequences siggraph signal sliding society spie stairway standard street synthesis synthetic system taylor technique technology tensors texture three tracking trifocal uncalibrated vcip vergauwen video vision visual wiegrand with zisserman zoom
http://csdl.computer.org/comp/proceedings/cvpr/2004/2158/01/215810652abs.htm	87	David Nister´	about accumulated accuracy accurate adverse aerial affected algorithm algorithms allow also alvey analysis angle angles apparent appear application approximately assoc authors automated automatic autonomous avoidance bauckhage behavior between biased bolles bottom boujou build buildings cambridge camera cameras cartography cases causally chiuso closed coherent column combined combining commun comp complete computation computer computing conclusions conditions conf conference consensus contained contract control corner cornering corp correspondence cvpr darpa data davison degrees dense depend despite detection detector detectors deviation discrepancy distribution document dominant dropouts dulong during edge efficient endorsements error errors estimates estimation european evaluation except exhibits expressed extended favaro features field figure fischler fitting fitzgibbon five flannery fourth frame from front further genc generalised geometry good gool government ground guide gutman haralick harris hartley having heading hierarchy hill http ieee illustrates image implementation implied institute integrated intel intelligence interaction interest international interpreted intl isbn journal julier keith left live lncs localization loops mach machine magnitudes maintain mapping maps march mcgraw meadow mean merge minimal mobility model mohr most motion multiple navigation nister note number numerical observe obstacle obstacles odometry official oliensis open operation other ottenberg over overlaid paradigm parameterization pattern people perception perceptor periods perspective photograph planes plotted point policies pollefeys pose poses preemptive press problem proc proceedings random ransac readily real recipes recognition reconstruction recovered recovery references relative representing resulting retained review right road robust royal sample schmid separate sequences should shown shows simultaneous since single site slip small soatto society solution solutions speed sponsored springer standard stephens stereo structure subdegree submitted successful such suggests supplement surviving system systems table technology tensor tensors terrain teukolsky that these thesis this those three time tomasi torr track traditional transactions trifocal uncalibrated under uneven university unmanned used vehicle vehicles verbiest verlag vetterling video view views vision visionbased visual volume were wheel which will with woods work world yields zero zisserman
http://csdl.computer.org/comp/proceedings/cvpr/2004/2158/01/215810536abs.htm	71	Clear Underwater Vision Yoav Y. Schechner and Nir Karpel Dept. of Electrical Engineering Technion - Israel Inst. Technology Haifa 32000, ISRAEL yoav@ee.technion.ac.il , karpeln@tx.technion.ac.il	academic adelson ahlen algorithm alon analysis angle animals aotf aperture approach aquatic archeology atmosphere atmospheric autom automatic autonomous ballard biol biological biologically bmbf bolas bottom boult caldwell cambridge camera capacity center chang chen cheroske chiou chitwood clear color combining comp comperative components computer concentration condiitons conf conference confocal congitive containing contrast correction cozman cronin cvpr cyber degraded degredation degree denes dept depth design diego difference directional dolphin dove effects elect emprical engheta engin enhanced equipment experim experiment fang farid fellow field flat flitton follow foresti foundation fountain from function funded gottlieb guichard harsdorf haze henry highlight home hopcraft horowitz horv http hull ieee ikeuchi illu image images imaging improving independent influence inspection inspired integrative intensity israel jaffe jakeman jordan josa kahanov kaminsky karpel kashiwagi keeping khong kiryati krotkov lagoon lavest lett levoy light lighting lightness linear linearly link mahadev marine marshall mcdowall mcglamery measurement media metes minerva mobley model modeling morgan multiview narasimhan natural nature nautical navigation nayar negahdaripour neuroscience nostrand oakley objects observation oceanic oceans ollendorff omni optical optics optimal orientations page pami part partial passive pattern patterns perception percieved photography phusical point polarimeter polarimetric polarization polarized poor portable press princeton proc proceedings process processing pugh quality radiative realtime recognition reconstruction references reflectance reflection reflections reflector refractionpolarization remains remote research response restoration results reuter robotics role rousseau rowe royal saito sato scatterer scattering scenes schechner scia semi sensing sensors sensory separating separation shallow shamir shashar shrimp shurcliff siggraph signalling signature simple sions skylight snell society somekh spatial spie spread state station statistical structures submersible sundgren supported surface surray synthetic syst system systems talyor tantura targets taub technion technique temporal testing through throught tran trans transfer transparent trial tropical understanding underwater uniform university urquijo using vaish variation varj vector vehicle vessel video visibility vision visual voss walker water waters weather webster wehner window with wolff work yoav
http://csdl.computer.org/comp/proceedings/cvpr/2004/2158/01/215810504abs.htm	67	L minimization in geometric reconstruction problems	appear applications based cambridge chirality computer conference criminisi cvpr environments exploration flannery geometry graphics hartley high http ieee image interactive international journal kang loqo manual multiple november numerical orfe pattern press princeton proceedings quality real recipes recognition references report society sturm szeliski technical teukolsky triangulation understanding university users uyttendaele vanderbei vetterling view vision winder world zisserman
http://csdl.computer.org/comp/proceedings/cvpr/2004/2158/01/215810560abs.htm	74	A Minimal Solution to the Generalised 3-Point Pose Problem David Nister´ Sarnoff Corporation CN5300, Princeton, NJ 08530, USA dnister@sarnoff.com Abstract	adelson aloimonos alter analysis anwendungen application arbitrarily archiv army assoc authors automated baker base based below bergen bolles calibration cambridge camera cameras cartography case classical commun comp computational computed computer conclusion conclusions conference configurations consensus contained cvpr detected dimensional distribution document early efficient either elements eral error erweiterter estimation euclid example explicitly expressed eyes facing faugeras fermuller figure finding fischler fitting five frobenius from function genclassical general generalisation generalised geodasi geometric geometry gestalt government grossberg gruen grunert grunerts haralick hartley have huang ieee image imaging implied intelligence international interpreted into isbn journal laboratory live lncs logarithm mach machine many mathematik method mirror model models motion multiple nayar nebst niste nolle norm numerical occurences occurrences official orientation ottenberg paradigm parameters pattern perspective physik plenoptic pless point points policies pose pothenotische preemptive presented press problem proceedings processing random ransac rays recognition references relative removes representing research restrictions review sample scale seine shaped should smile society solution solutions springer structure such terms this those three transactions trials type uber unit university using verlag view viewpoint views vision visual volume weak whereas which with works zisserman
http://csdl.computer.org/comp/proceedings/cvpr/2004/2158/01/215810462abs.htm	62	Using Skew Gabor Filter in Source Signal Separation and Local Spectral Multi-Orientation Analysis	acoustics adelson advantages after also america analysis analytically applications approximatation attractive barron based beauchemin bergen book boundaries bovik bracewell browsing cells clark communication company complete complex compression computational computer conference correcting cortex cvpr data daugman deblur describing difficult discrete distribution eigenenergy energy estimate eventually farid farrokhnia features filtered filters flow fourier framework frequency gabor gaussians geisler girosi grzywacz heeger higher hill ieee image intelligence international jahne jain journal learning limited local localized london machine manjunath mase mcgraw model models moment more motion multichannel networks neural nevertheless occluding optical order orientation overcome pages pattern perception poggio possible proc proceedings processing property recognition recognized references representation representations resolution retrieval royal segmentation shizawa should shows signal signals simoncelli simplicity skewness society spatial spatio spatiotemporal spectral speech springerverlag statistical steerable straightforward structure temporal texture theory trans transform transforms transparency unified unsupervised using velocity very vision visual wavelets wedge while without yuille
http://csdl.computer.org/comp/proceedings/cvpr/2004/2158/01/215810255abs.htm	34	Multi-Classifier Framework for Atlas-Based Image Segmentation Torsten Rohlfing and Calvin R. Maurer, Jr. Image Guidance Laboratories, Department of Neurosurgery, Stanford University Stanford, CA 94305-5327 rohlfing@stanford.edu, calvin.maurer@igl.stanford.edu	alignment ambleside anal analysis animal application applications approach artif atlas automatic bagging bartels based bees bellingham berlin biomed biomedical boca brain brains brandt breast breasts breiman chapter classifiers collignon collins combination combining comput computer conference confocal cvpr cybern dawant deformation deformations demaerel dimensional duin entropy environments evaluation evans expectation experts fischer fitzpatrick flexible flirt form forsey free front geometric graph hajnal handbook handwriting handwritten hartmann hatef hawkes hayes head heidelberg hierarchical hill holmes huang ieee imag image imagery images imaging inform information intell intern internal international invariant ipmi july june kichenassamy kittler krzyzak kumar label leach learn lecture level mach machine maes maintz malladi marchal matas maurer maximisation maximization measure medical memory menzel method methodology methods microscopy model modelling models modersitzki multi multimodality multiple multiprocessor mutual neuroanatomical neuroimage noble nonlinear nonrigid normal notes numerals olver overlap papers parry part pattern peters philadelphia predictors press proceedings processing propagation raton recognit recognition references refinement registration revised rigid rohlfing rueckert russakoff science second sederberg segmentation selection sethian shape shared similarity snake society solid sonka sonoda spie spline springer strategies structures studholme subjects suen suetens syst tannenbaum taylor technol their thirion three toolbox trans transformations unconstrained using validation vandermeulen vannier vemuri verlag vision volume wbir with workshop yezzi
http://csdl.computer.org/comp/proceedings/cvpr/2004/2158/01/215810568abs.htm	75	Stereo correspondence with slanted surfaces: critical implications of horizontal slant Abhijit S. Ogale and Yiannis Aloimonos Center for Automation Research, University of Maryland, College Park, MD 20742 {ogale, yiannis}@cfar.umd.edu	adaptive algorithm algorithms allowing also approaches approximate april barnard based baseline bayesian between binocular birchfield black bobick bottom boykov cameras comparisons computation computational computer computing conclusions conference constraint constraints correspondence correspondences cuts cvpr daniilidis dense detecting diffusion discussed disparity dissimilarity distributions dynamic eccv effects efficient egnal empirical energy enforce evaluation existing experiment experimental fast figure finally finding five flow formulation frame framework from furthermore fusiello geiger geman gibbs global graph halfocclusions have hence horizontal horizontally human iccv ieee ijcv image images implements imposes insensitive instead intensity inter intervals intille intra july june kanade kolmogorov kumar ladendorf large layered leads lecture left level local london march marr matching maximum measure methods metrics minimization modeling modified modify motion mulligan multiple multiway must ncamera nonlinear notes object occlusion occlusions ohta okutomi over pages pami pattern pixel pixels poggio predicting presented problem proc proceedings programming projections provided real recognition references relaxation requires restoration results right roberto royal sampling sawhney scale scanline scharstein science search sept shown simple slant slanted society stereo stochastic suitably surface surfaces szeliski taxonomy that theory these time tomasi trans trucco uncertainty unequal uneven uniqueness using veksler vision visual which wildes window windowing windows with yuille zabih
http://csdl.computer.org/comp/proceedings/cvpr/2004/2158/01/215810754abs.htm	100	Motion Estimation by Swendsen-Wang Cuts	about accreted accurate algorithm algorithms allows also analysis approach assumptions barbu based bayesian belief binocular black carlo chain coherence collective combine combining comparison competition computational computer concept conference consider correspondence cremers current curvature cuts cvpr data deal deformations dense dependent deterministic discontinuities distribution doretto driven dynamic eccv encoding estimating estimation evaluation flow frame framework france freeman from fronts gaucher geiger geman geometric gibbs girosi graph grzywacz horn iccv identical ieee ijcv image images implementation jepson ladendorf layers lett local made make markov mathematical maximize medioni mixture model modeling models monte more motion motivated mrfs nice nonparametric number occlusions only optical order osher pami paper parallel parameters parametric particle particular partition pattern philosophical phys physics posterior press problem proc proceedings proof propagating propagation provide realistic recognition reconstruction references region regions regularization relaxation restoration robot royal scharstein second segmentation segmented selection sequence sethian shape simplifying sketch smoothness soatto society speed spin stage stereo stochastic surface swendsen swendsenwang systems szeliski tappen taxonomy texture textured them theory these this three tion torr trans updating using variable variational velocities visible vision wang wave weiss which with wolff work yuille
http://csdl.computer.org/comp/proceedings/cvpr/2004/2158/01/215810478abs.htm	64	Separating Style and Content on a Nonlinear Manifold Ahmed Elgammal and Chan-Su Lee Department of Computer Science, Rutgers University, New Brunswick, NJ, USA {elgammal,chansu}@cs.rutgers.edu	able acknowledgment active activity addressed advances algorithm also america analysis appearance application applications approach approximation award based being belhumeur between bilinear body both bottom brand bregler calculus chui class classification clustering coding cognitive component computation computer conference content continuous cooper coordinates cootes copenhagen cviu cvpr danmark decomposing denmark different differential dimensional dimensionality dynamic eccv econometrics eigenfaces elgammal embedding ensebles expression extracting facial factor figure fisherfaces four framework freeman from functions funded gait girosi graham half handle handles hawai hespanha huang human iccv ieee illumination image important improves inferring information input interesting intermediate international interpolated intersection john jolliffe journal july june kapteyn kriegman last learn learned learning left levin linear locally magnus manifold manifolds mapping maps marimont matching mathematical matrix mention mode model models motion multilinear multiple murase nayar network neudecker neural neuroscience ninth nonlinear notes objects observation observations omohundro optical over page pages paper parameters parametric partially past pattern pentland people perceptual poggio point pose principal principle proc procedure proceedings processing projection proposed psychometrika puppetry question rangarajan rank recognition reduction references regression research rigid roweis saul sciene separate separating shadow shape shashua shown signatures silhouettes simple smile smiles society some sons space spaces specific spectra spectral speech springer statistics style styles subspaces surface taylor tenenbaum tensor tensorfaces terzopoulos that their theorem this three time training tucker turk unified unifying unsupervised using vasilescu verlag vision visual volume walk walks wandell wansbeek wiley with work workshop yields york
http://csdl.computer.org/comp/proceedings/cvpr/2004/2158/01/215810024abs.htm	3	Fast Wide Baseline Matching for Visual Navigation Toon Goedeme´1, Tinne Tuytelaars1, and Luc Van Gool1 [toon.goedeme, tinne.tuytelaars, luc.vangool]@esat.kuleuven.ac.be	across affine affinely algorithm analysis appearance applications approximate arya assoc automated automatic auton based baseline bauckhage baumberg biomimetic bmvc bolles carlsson cartography chum civr color commun comp computed computer conference consensus control cvpr database detector dissertation doctoral eccv ethz extremal fast feature features fischler fitting franz from goedeme gool greyvalue hpat html http icassp iccv ieee illumination image indexing inertial intelligent intensity interest invariant invariants irrespective journal ligtenberg line local loeffler lowe mach mallot matas matching maximally mikolajczyk mindru model mohr moons moschytz mount multiplications navigation nearest neighbor netanyahu nuttin object optimal pajdla pami paradigm pattern patters point practical proceedings profiles random recognition recognizing references regions reliable retrieval robot robotic robotics robust role sampling scale scene schmid searching sensing separated servoing shao showroom silverman society stable stereo stockholm svoboda systems tell topological tuytelaars urban using viewpoint views vision visual wheelchair wide widely with zisserman zubud
http://csdl.computer.org/comp/proceedings/cvpr/2004/2158/01/215810790abs.htm	105	Multiple Kernel Tracking with SSD	academic acknowledgements adapt added addition algebra algorithms alignment always anal annals appearance applications applied apply approach area associated asymptotically authors baker based becomes belhumeur best between bhattacharyya blob brute centroids challenging changing chapman cheng circularly class clear clustering collection collins comaniciu communication comp computational computer conf conference connection constrained constraints continuing continuous current cvpr decision detecting detection develop developed discussions distance divergence doing edge edition efficiency efficient emphasizes equivalence estimating estimation example extreme extremes features finally first force forge formal foun fundamental gain geometry good hager hall hand helpful histogram histograms hope ieee illumination image images insights integral intelligence itself john jones kailath kernel kernels like likewise limit linear link mach make making material mathematical matthews matusita mean measures meer methods mode models more motion motions multiple national november object occlusion october only optimal optimization orlitsky other over pages pami parametric patt pattern places press probability problems proc proceedings properly ramesh recog recognition recover references region registration relationship results rich robustness rules same samples santhanam scale science search second seeking segmented selection sense shift signal similar simple simultaneously smoothing society solved space spatial statistical statistics strang structure such supported symmetric target techniques technology template thank theory there these this through time tomasi track tracking trans transactions turing typically upon using vision wand well while with work would xiangtian zhang
http://csdl.computer.org/comp/proceedings/cvpr/2004/2158/01/215810247abs.htm	33	Segmentation Using Multiscale Cues Stella X. Yu Computer Science Division University of California at Berkeley Berkeley, CA 94720-1776, USA stellayu@cs.berkeley.edu	about achieved acknowledgments across affinity algorithms allow also alternative altogether america analysis andras andrea application approach approximations associated assumed author average away based basri belongie best better body boundaries boundary brandt brightness canny care carnegie choosing clustering coarse combining comm comments complete completion computation computational computer conclude conference configurations connections contain continuity contour contours contribute converge cues curvy cuts cvpr database dence detect detection deva discrete discrimination early easier ecological edge efficient efros eigensolution emerge entries erik european evaluating extreme factors fast faster ferencz fields figure finally financial follow fowlkes france friends from frome functions further future grant graphs grouping have here human iccv ieee illusory illustrate image images information intelligence international jacobs jitendra journal just keeps learned learning leung level like lindeberg linear local long machine make makes malik martin math matlab measuring mechanisms mellon memory miller model models mori most motivating multi multiclass multigrid multiscale mumford muri natural neighbours neural nice nonzero normalized number numerically obtained often only optical optimal organization other pages paper parallel pattern perceptual perona piecewise pixels preattentive probabilistic problems proceedings processing pure ramanan range ranges recognition recovering reduction references region reliable research respect respectively results running scale scales segmentation segmentations segmented shah sharon short signals simple size smooth society space sparse spectral speed statistics stochastic structures suggests support systems tail take technique texture thanks that their them thesis they thin this those time together torso total transactions university using variational vision weak whereas williams with within work
http://csdl.computer.org/comp/proceedings/cvpr/2004/2158/01/215810261abs.htm	35	Efficient Belief Propagation for Early Vision Pedro F. Felzenszwalb and Daniel P. Huttenlocher Department of Computer Science Cornell University {pff,dph}@cs.cornell.edu	accuracy adelson advances ages algorithm algorithmic algorithms analysis anandan another appear applicable applications applied apply approach approximate arbitrary based because been being belief bipartite black borgefors both boundary boykov broad burt code commonly communication compact comparison computer conference constant convolution correspondence cost costs cuts cviu cvpr decimation decrease dense diameter differs digital directly discontinuity disparity distance energy estimates estimation evaluation evidence experiments extension extensions factor fast felzenszwalb fields flow frame freeman from functions gaussian graph graphics graphs grid here hierarchical hierarchies high hmms however huttenlocher identical ieee image important including information intelligence international iterations journal kleinberg laplacian large longer lower machine markov matching memory mentioned message messages method methods minimization models more motion motions multiple multiresolution namely neural number object obtain occlusion only opposed optimality parameters parametric particularly passing pattern pictorial piecewise pixel possible problems proceedings processing product propagation proportional pyramid range rather recognition reduce references related relatively reported requirements resolution robust running scharstein section shown shum signal simplest small smooth society solutions solve sophisticated space state stereo storage structure structures such systems szeliski tappen taxonomy technique techniques than that theory there third this three time transactions transform transformations usage used uses using vary veksler vision weiss were willsky with would zabih zheng
http://csdl.computer.org/comp/proceedings/cvpr/2004/2158/01/215810452abs.htm	61	Making One Object Look Like Another: Controlling Appearance Using a Projector-Camera System	adapted allows anaheim angle animating animation annual apparent appeance appearance appears approach automatic background balancing ball bandyopadhyay based being belhumeur blending both brand brick brightly calibration camera cameras cartoon cave cham change changed chen color colored columbia combining compensation computer conference continuous controlling cruz cutts cvpr debevec defanti department design desired devlopment differently digital dioramas display displays distributed dome dynamic edge electro elimination environments eurographics experimental experimentalresultsfromavideoshowingtheglobewith exposed extending field fields fifth figure form forum from front fuchs funayama future gamut gives globe graphics grossberg hanrahan haptic head here high humphreys ieee illumination image imaging immersive imperfections implementation inami inova inter international into itec japan june jupiter kawakami lake lamps large lasers light lyon maeda majumder malik mann maps matching mitsunaga modeling motion mounted multi multiple nayar neira npar number objects occluder office optical optics optique pages patent pattern peri photographs photorealistic picard pictures playing press preview proc proceedings projected projection projector radiance radiometric range raskar real reality recognition recovering references rehg rendering report reproduction results rotating sandin scene science scottsdale screen seamless sekiguchi self shader shadow side siggraph soccer society space spatially stesin stone sukthankar superposition suppression surface surround symp system systems tachi tajima technical technology that this tiled transformed tsukada uncompensated undigital unified university using video view views virtual vision visualization visuo volume wall wallace walls welch wide willwacher with workshop yanagida york ziegler
http://csdl.computer.org/comp/proceedings/cvpr/2004/2158/01/215810856abs.htm	114	Modeling Complex Motion by Tracking and Editing Hidden Markov Graphs	advances algorithm algorithms animation appear appearance applications approximate approximation april based blau brand bregler buffer calculations canada character chemical combining company computation computer computing conference copenhagen cvgip cvpr data denmark dept detection digital distance doretto dynamic ebert eccv edit efros elements entropy equations essa estimation evaluation experts fast fitzgibbon flexible forms france freeman gaseous graphics haralick hertzmann iccv icip ieee ijcv image images implementation incomplete information interactions joseph julesz july kimia klein lausanne learning leung level levoy lischinski lopez lumbreras machines malik marr matching math mathematical mcgill mcmc method methods metropolis minimax mixing mixtures model modeling models moment motion movie mumford nature neural nice nowhere object pami parametric parent particle pattern perception phenomena physics picard preprint primal principle probabilistic problems proc proceedings processing quantization recognition reeves references registration rendering ridge ridges ridigity rosenbluth salesin sampling scaleline schodl sebastian second senes serrat shading shape shum siggraph sketch sketchability soatto society soda stat state static statistical stochastic structured style switzerland synthesis system systems szeliski szummer techniques teller temporal textons texture textured theory theri trans tree univ using valley valleys vancouver vector video villanueva vision visualization volume wang wave werman what with yaniv yuille
http://csdl.computer.org/comp/proceedings/cvpr/2004/2158/01/215810415abs.htm	56	A Segmentation-Free Approach for Skeletonization of Gray-Scale Images via Anisotropic Vector Diffusion Zeyun Yu and Chandrajit Bajaj Department of Computer Science, University of Texas at Austin, Austin, Texas 78712, USA Emails: zeyun, bajaj @cs.utexas.edu	analysis aniso arcelli baja between bias blum brain bruckstein characters color comparison comprehensive computer computing conference cvpr data demonstration descriptors distance dunn edge efficient english euclidean extracting figure form grad graphics hierarchic ieee image intelligence kimmel kiryati kubler letters level machine malandain maps method methodologies models objects ogniewicz original pattern perception points press problem proceedings recognition references removal results ridge sets several shaked shape skeleton skeletonization skeletons smth society speech strength suen survey tests thinning toga trans transactions transformation understanding vidal vision visual visualization volumetric voronoi wathen zhou
http://csdl.computer.org/comp/proceedings/cvpr/2004/2158/01/215810364abs.htm	49	Video Repairing: Inference of Foreground and Background Under Severe Occlusion Jiaya Jia Tai-Pang Wu Yu-Wing Tai Chi-Keung Tang Vision and Graphics Group, Computer Science Department Hong Kong University of Science and Technology {leojia, yuwing, pang, cktang}@cs.ust.hk	about achieve adaptive addison adelson after agarwal alignment amsderstam analysis anandan annual applications approach arno baker ballester barrett based because been belongie bertalmio black blake built camera caselles caspi chairs charles chroma class coherent color comaniciu composition computational computer conference consistency contour contributions cooperative creating cvpr damage damaged darrell david davis deformations deployed depth different dyer edge elseviers environment environments essa estimating estimation extraction feature figure fixed flow foreground frames framework from full future gangnet generation gortler graphics handled heung hexoctahedral iccv ieee image images including incorporating indexing inpainting input intelligent interactive interested international irani irfan jepson jetstream jiaya john jonathan josh keung knowledge large layer layered layers lengyel lighting local maintained maps march mean medioni meer mishima models more morphing mortensen mosaic mosaics motion motions movels moving much multiple objects occlusion october optical order pages pami panning panoramic parallax parametric particles pattern pentland perez performed pixels plausible point press probabilistic proceedings publishing ramesh real recognition reconstruction reference references rendering repaired repairing representation representations restoration result richard rigid robust rotation salesin sameer sample sapiro scenes schodl science scissors sculpture segmentation segmented seitz sequence serge severe shade shadow shift shum siggragh siggraph snyder society soft space spatial static step stereo steven support synthesis system szeliski tang technical techniques temporal tensor textures that these time towards tracking trans upon using variable velocities video view virtual vision visually voting wang well went wesley what where wills with without works yeung zoom zoomin
http://csdl.computer.org/comp/proceedings/cvpr/2004/2158/01/215810684abs.htm	91	Lie-Algebraic Averaging For Globally Consistent Motion Estimation	about affect algebras algorithm alignment along analysis angle annual applications automatic autonomous averages averaging avoid axis balanced based bergen buss calibration cambridge camera castle choose cipolla combining communication component computer computing conference conics consistent constraints construction could cross curtis curve cvpr darpa defence department depth determining direction engo environment environments epipolar espirit estimate estimated estimation euclidean european extent factorization figure fillmore first fitting fitzgibbon fletcher focal formula frame from geometrical geometry georgescu globally govindu graduate graphics group groups half hartley here ieee image images indicate indicates individual informatics inria international interpolation intrinsic invariance janin jiang joshi journal july kanade kanatani kang knowledge large least lecture length line lines magnitude mapping mathematics matrix means meer mendonca method methods metric milios moakher model motion multi multiple nonlinear note notes original orthography over parameters pattern pixels point points pollefeys positions press principal proceedings projective proportional quality quan rangescan reality recognition reconstruct reconstruction recovering recovery references report representation representations research respect result review robotics robots rotation rotations scale scales science self selig september sequence sequences shape should shows siam simple single size society some spherical spite splines springer squares stage statistics streams structure sturm symposium szeliski table technique terms texts their theoretical tomasi trajectory transactions triggs tsui turn uncalibrated uncertainty under understanding university unknown used using varadarajan varying verlag view viewed viewing views virtual vision visual which with workshop yaxis zhang zikan zisserman
http://csdl.computer.org/comp/proceedings/cvpr/2004/2158/01/215810811abs.htm	108	Model-based motion clustering using boosted mixture modeling Vladimir Pavlovic Dept. of Computer Science Rutgers University Piscataway, NJ 08854	action actions activity additive advances agrawal algorithm algorithms alon american analysis annals applications approach approaches approximation april artif aspect assignments august bartlett based baxter bayesian been belief biological bissacco biswas bobick boosted boosting bootstrap bottom brand bregler brief california cambridge carnegie chapman character classifiers clustering clusters clutter cohen combining complex comprised computer computing conf conference coupled cuzzolin cvpr data databases dataset debrunner demonstrated density depicts dept detection determined dimension discovering discovery distance distribution distributions durbin dynamic dynamics eddy editors efficient estimating estimation event faloutsos figure finite firoiu foundations four frean frezza friedman from function functional gaffney gaussian generalized gesture gradient greedy grimson group grouping hall hastie heckerman hidden hogg huber human humo husemeier hypotheses ieee ijcai image individual information international into introduction irani jaakkola james john johnson jordan journal juang kernel knowledge kollios krogh krose language large learned learning linear machine madison management manor margin markov mason mclachlan meek meila mellon methods microsoft mining mitchison mixture mixtures model modeling models motion mozer networks neural nips oates object oliver operations optimal organization pages pami panel paradigms parametric past pattern patterns pavlovic penny pentland periodic petsche press probabilistic proc proceedings processing rabiner ratio real recognition recognizing recursive references regression report research rezek roberts robust rosset sampled schapire schwartz sclaroff search segal segmentation selected september sequence sequences series shows shum siggraph sign similarity smyth snowbird soatto society sons southern sparsely speech springer staged stanford starner statistic statistical statistics stauffer stockholm subjects subspace sugar sure swami sweeden switching synthesis systems technical techniques temporal texture thiesson tibshirani time towards tracking tractable trajectories trajectory tree tutorial twolevel ucla university unsupervised using utility verbeek video vision vlassis volume wang warping whose wiley willey wilson with workshop zelnik
http://csdl.computer.org/comp/proceedings/cvpr/2004/2158/01/215810510abs.htm	68	A New GPCA Algorithm for Clustering Subspaces by Fitting, Differentiating and Dividing Polynomials	according algebraic also among analysis analyzers apperances applied applying approach assume because between bischof bishop both camera case cases cluster clustering collection component components computation computational computer conditions conference consider consisting coordinates correspond corresponding costeira could cvpr data database decomposition different differential dimension dimensional downsampled each eccv efficiency eigenspaces eigenvectors face faces factorization figure first fitted frames from frontal generalized given goes gpca grouping guest hartley hirsch homogeneous host huang iccv ieee ijcv illumination image independently instance international interviewing into kanade kanatani kriegman left leonardis lighting like linear live lives lower maver method missing mixtures model motion moving multibody multiple neural news number objects obtain obtained onto origin original orthogonal pages pattern perfect perspective pixels points powerfactorization principal probabilistic proceedings projected projection recognition references results right rows same sastry scene scenes segment segmentation selection separation sequence sequences shown shows small society space springer subjects subset subspace subspaces switching tensor that them then three through tipping topology trifocal under unified unifying using varying vidal video view views vision volume weiss which with would yale yang zhang
http://csdl.computer.org/comp/proceedings/cvpr/2004/2158/01/215810234abs.htm	31	Color Alignment in Texture Mapping of Images under Point Light Source and General Lighting Condition	absolute accv albedo albedos asian attribute automated automatic automation beauchesne between both bottom color computer conf conference constrained cvpr data difference donald estimated eurographics fifth figure finkelstein from fusion geometric gortler graph graphics greenberg heidrich hoppe ieee ikeuchi image images imaging input inverse irradiance klein kurazume lapped lensch levy lighting mapping marschner meshes model models multiple neugebauer nishino note objects overlapping pacific pattern photographic photography polygonal praum proceedings progressive pseudo real recognition references reflectance registration relighting robotics sander seidel siggraph simultaneous snyder society stephen stitching synthetic test texture textures texturing that unregistered using utilizing views vision wheeler workshop world zhang
http://csdl.computer.org/comp/proceedings/cvpr/2004/2158/01/215810326abs.htm	44	Similarity Measure and Learning with Gray Level Aura Matrices (GLAM) for Texture Image Retrieval	abbadeni accurate acknowledge acknowledgments adapted advances aggarwal alberta algorithm algorithms alignment also analysis anonymous appear application applications approach aura authors azencott ballard based bayesian behavior bergen between bonet bottom browsing buhmann burges calculated calculating cambridge chan chang chapter classification cluster coefficients color combined comments complex computer computing conclusion conf conference content cooccurence cooccurrence cooccurrences copeland corner correctly cross cvpr data databases davis defined department discovery discussed discussions dissimilarity distance distances distributions divergence each eccv edition effects elfadel empirical engineering entropy evaluation example examples existing experimental explain face fast feature features feedback field fields figure filters finally first fourier frame from function funding gamma geman generalized generated generous gibbs given gives glam glams graphics gray grayscale group heeger hofmann however iapr iccv icpr ieee image images imaging incorrectly indexing information intelligent interactive international intl issue jain joachims john johns joint journal kalto kernel kullback labeled large learning least left leibler level levoy like located lohmann making manjunath markov matching mathematical mathiassen matrices matrix maximum measure measures measuring members method methods minka miscibility model modeling models more multi multimedia multiresolution mumford natural networks neural nonparametric nserc obtained occurrence optical order ordered orengo pacific pami paper parametric patt pattern patterns pentland perceptual perceptually photorealistic picard picture popat portilla pose practical present presented press probability proc procedure proceedings processing psychophysical puzicha pyramid pyramids quantization query random rank ranked ranking ravichandran recog recognition references relaxation relevance rendering restoration results retrieval retrieved retrieving reviewers robots sampling santini scale scholarship science section segmentation shiftable siggraph signed similar similarity simoncelli skavhaug smola society sons special spie statistical statistics stochastic storage stricker structure sung support swain synthesis systems temperature tetxure texture textures than thank that their theory they this towards trans transactions transforms treestructured trivedi under unified university unsupervised using vapnik varing vector video viola vision wang wavelet wavelets well which while wiley windowed with workshop would xture york younes
http://csdl.computer.org/comp/proceedings/cvpr/2004/2158/01/215810421abs.htm	57	Tracking Loose-limbed People Leonid Sigal Sidharth Bhatia Stefan Roth Michael J. Black	adaptive annealed approach articulated assembling attractive automatic background bayesian belief black blake body boosted bottom bregler burl calibration camera capture cardboard carlo cascade cham computer computers conf conference coughlan covariance cvpr deformable detection deutscher doucet dynamic eccv efficient elschlager entropy exponential face features felzenszwalb ferreira field figure figures filtering filters finding fischler fleet forsyth four frame frames freeman freitas from further gaussian geometry gesture gibbs global gordon graphical grimson hand human huttenlocher iccv ieee ihler ijcv image images info initialization interface ioffe isard jones learned learning likelihoods limbed local loopy loose maccormick malik manhattan maps markov matching maximum methods mixture mixtures model modeling models moghaddam monocular monte motion multiscale mumford murphy network nips nonparametric object objects pami pampas parameterized parametric particle partitioned pattern pavolvic pentland people perona photometry pictorial practice probabilistic proceedings products propagation quality ramanan random rapid real recognition references rehg reid representation roth sample sampling scaled sciences sequential shapes shown sidenbladh sigal sigelman simple single sminchisescu society springer statistics stats stauffer stochastic structures sudderth texture theory time towards tracked tracking trans trees triggs twists unified used using valued verlag video views viola vision visual weber whose willsky with yacoob
http://csdl.computer.org/comp/proceedings/cvpr/2004/2158/01/215810842abs.htm	112	1	alexander analysis apparent arridge august barker basser bihan brain coefficient comp computer conference cviu cvpr data dept detection diffusion diffusiontensor echo effective effects elastic elucidated estimation features from gaussian hasan human ieee images indices lazar magnetic matching mattiello microstructural modeling note notes october parker partial pattern physiological pierpaoli proceedings profiles quantitative recognition references research resonance self shape similarity society spatial spin tensor tensors tissues transformations tsuruda vision volume
http://csdl.computer.org/comp/proceedings/cvpr/2004/2158/01/215810894abs.htm	119	Articulated Models from Video	absolute acquired acquisition acqusition across adjustment alan algorithm alignment amer analysis annealed appearance applications appuzzo articulated automated automatic available baker based bayesian black blake body boston bottino bregler bundle calibration canada capture center cheung clipping closed combining computer conf conference connected considered consisting corfu covariance crisis cvpr data david december deutscher documentation exponential figure filtering first forsyth from fusion generic greece heads hedvig horn http hull human ieee image initialization intelligence interaction international introducing ioffe jenkins joints journal july june kakadiaris kanade kettebekov keyframes kinematic krahnstoever laurentini learning library links machine madison malik management maps markerless marriott mataric metaxas michael mixtures model modeling models monocular motion multimodal multiple murta nils november orientation orlando overlaid park particle pattern pennsylvania people plaenkers polygon pond positions problem proc proceedings quaternions recognition reconstruction references refinement regularized reid relative released sampling scaled schapira september sequences sergey shape sharma shown sidenbladh silhouette simon sminchisescu society software solution sources staff state statistics stereo systems takeo thesis three time tracking transactions trees triggs twists unit university unknown using vancouver video viewpoints views vision visual volume wacv when wisconsin with without workshop world yeasin
http://csdl.computer.org/comp/proceedings/cvpr/2004/2158/01/215810488abs.htm	65	Video Data Mining Using Configurations of Viewpoint Invariant Regions Josef Sivic and Andrew Zisserman Robotics Research Group, Department of Engineering Science University of Oxford http://www.robots.ox.ac.uk/vgg Abstract	affine affinely agarwal analysis aner appearance applied approach automatic based baseline bmvc carlsson cars cast chum class cluster clustering combining computer conf conference constraints constructing content correspondences cvpr data descriptors detection detector discovery discrete distinguished eccv eickeler expo extremal face faces features fergus ferrari finding fitzgibbon foundations frames from generating gong google gool graphics greyvalue groups guide hamilton hilderman holiday hong huang icassp iccv ieee image images indexing inexact interest international intl introduction invariant invariants iurgel john journal kanade kaufman kender kluwer knowledge language lazebnik learning lienhart listing lncs local lowe manning matas matching mathematics maximally measures method methods mikolajczyk mining mmsp modeling models mohr mosaic movies multi multimedia multipleview multiview natural neighborhood obdrzalek object optimal organize pages pajdla pami patches pattern patterns perona personalization point points ponce practitioner press proc proceedings processing recognition references regions reliable representation retrieval rigoll robust roth rothganger rousseeuw scale scene schaffalitzky schmid schneiderman schutze sets shot sivic smith snaps society sons sparse spatial springer stable statistical statistics stereo summaries summarization survey system tell text texture through topology transition tseng tutelaars tuytelaars unordered unsupervised urban using verlag video videos view vision volume wallhoff wide wiley workshop zisserman
http://csdl.computer.org/comp/proceedings/cvpr/2004/2158/01/215810074abs.htm	10	Segment-based Stereo Matching Using Graph Cuts Li Hong George Chen Advanced System Technology San Diego Lab, STMicroelectronics, Inc. li.hong@st.com george-qian.chen@st.com	acknowledgement adaptive algorithm algorithms analysis approximate areabased ball based bdisc belief birchfield boykov buntex camera collective color comaniciu computation computer computing conf conference contours cooperative cornell correspondence corresponding corresponds cuts cvpr data dense detailed detection different diffusion disparity each edge efficient energy estimation european evaluation evolutionary extracted fast feature figure fitting frame framework from functions garden global goulermas graph grouped ieee image images improve index initial integral intelligence intermediate isprs july kanade kolmogorov kumar labeling layered liatsis like local machine maps matching mayer means meer methods middlebury minimization minimized model motion multi multiway nonlinear novel occluded occlusion occlusions pattern photogrammetric pixels plane proc procedure procedures proceedings prof propagation providing purple recognition reconstruction reference references refined regions results robust sawhney sawtooth scene scharstein segment segmentation segments sept shum single slanted society spaces stanford stereo studio surface surfaces symbiotic szeliski table taxonomy testing thank their thesis tomasi trans tsukuba university using variable veksler venus views vision visual website what window with would zabih zheng zitnick
http://csdl.computer.org/comp/proceedings/cvpr/2004/2158/01/215810848abs.htm	113	Affine Image Registration Using A New Information Metric	alignment amini analysis assisted automated ayache based benavente biomedical boes bookstein brain case clustering colchester collignon collins computer computing conf conference connected construction control cover culty cvpr database design digital discrete distributions elements entropy evans examine extending face feature fifth from genant hanson hawkes hawkins hill holmes iccv ieee imag image images imaging incorporating information intermodality intervension intl into invariant john kabani knee kollokian labelling lecture lynch maintz marchal martinez mathematical maximization measure medical medicine methods metric meyer modality multi multimodality multiple mutual notes osteoarthritis overlap pattern peterfy phantom pluim press probability proc proceedings processing pulse rajski realistic reality recognit recognition references region registration report robotics science sequences sled society sons space spect spie springer studholme suetens survey taylor technical theory thomas trans using vandermeulen variate verlag viergever viola virtual vision wells white wiley wilson york zijdenbos
http://csdl.computer.org/comp/proceedings/cvpr/2004/2158/01/215810668abs.htm	89	Non-Rigid Shape and Motion Recovery: Degenerate Deformations Jing Xiao Takeo Kanade The Robotics Institute Carnegie Mellon University Pittsburgh, PA 15213 {jxiao, tk}@cs.cmu.edu	affine alexander algorithms alignment analysis antipenumbra applications area avidan baker biermann blanz body brand bregler calibration camera cast chai closed computer computing conf conference constraints correction costeira cvpr dynamic eccv efficiency equivalence euro faces factorization form from fundamental gool hartley hertzmann heyden huynh ieee image independently intelligence internal journal kanade koch light linearly machine matrices matrix matthews method methods metric model modeling models monocular morphable motion moving multibody multiple nonrigid objects orthography outlier parameters paraperspective pattern perspective poelman points pollefeys proc proceedings projection projective quence rank recognition reconstruction recovering recovery references rigid sastry scene scenes segmentation self sequences shape shashua siggraph soatto society solution source spite streams structure synthesis teller their tomasi torresani tracking trajectory trans triangulation triggs under unknown varying vetter vidal video views vision with wolf workshop xiao yang
http://csdl.computer.org/comp/proceedings/cvpr/2004/2158/01/215810707abs.htm	94	Multibody Factorization with Uncertainty and Missing Data Using the EM Algorithm Amit Gruber and Yair Weiss School of Computer Science and Engineering The Hebrew University of Jerusalem Jerusalem, Israel 91904 {amitg,yweiss}@cs.huji.ac.il Abstract	algorithm algorithms amit analysis anandan angular applications applied applying around axes axis body bottom cans cations characterizing classification clean coaxial coherence compared comparison computer concentric conf conference consistency consisting correct costeira cube cvpr cylinder cylinders data degeneracy details deviation different display displayed dynamic eccv editor elongated entries errors estimation evaluation existing experiments exploiting factor factorization fail feng figure first found frame from function fundamental further gear gelb graphs grouping gruber gure iccv ieee ijcv image images input inputs intensity interpertation irani jacobs january kanade kanatani level line linear machline manor matrix method misclassi missing modeling models morris motion multi multibody nips noise noisy november number object objects observation obtained only optimal orthography other overlap pages parallel pattern percentage performance perona perpendicular perspective planes plus points press proceedings psychometrika real recognition reconstruction references result results revisiting right rotating roweis rubin same sastry scene scenes second segment segmentation segments selection sequence sequences shape shashua show shown sigma similar soatto society software some spca speed speeds standard streams structure succeeds superimposed synthetic temporal text thayer their time tomasi total tracking tracks truth tting ullman uncertain uncertainty under velocities versions vidal view views vision visual weiss when while with without wolf workshop yair zelnik
http://csdl.computer.org/comp/proceedings/cvpr/2004/2158/01/215810880abs.htm	117	3D Facial Tracking from Corrupted Movie Sequences	academic accv acknowledgments active adaptive adjust agreements algorithm animating animation application applications approach athitsos august based bayesian because being bhotika biermann biomedical bishop blake blanz blatantly brand brasileiro bregler brown bundle burgard capes capturing carlo cascia challenge chapter chung cnpq coding compression computer condensation conditional conference constraints contours control cooperative cootes course currently cviu cvpr deformable dellaert density does dynamic eccv edge edition either elements estimation experiments face faces facial faep fapesp fast figueiredo filter filtering finite flexible flow forsyth from funds gallaudet goldenstein good gordon graphics have head hertzmann huang iccv ieee ijcv illumination image images imaging impa institute integration interdisciplinary introduction investigating isard july kalman king kluwer level liability localization make mapped matching matem maybeck medical meer ment merwe metaxas methods mobile model modeling models morphable motion motionguided motions nasa national networks neural nongaussian nonlinear nonrigid notes novel number numerical optical outlier pages pami paper parameters pattern physics pighin point press probability proc procedures proceedings processing propagation publishers publishing radar realistic recognition recovering recovery references registration rejection relaxing reliable research restoration resynthesizing rigid robot robust salesin salmon samples scientist sclaroff self shan shape shapes siggraph signal smith society space springerverlag state statistical statistics stewart stochastic stolfi streams subtle supported synthesis system szeliski taylor techniques template texture texturemapping that their theory this through thrun tracker tracking training trans turned tyler under unfortunately unicamp unscented using validated varying vetter video violated vision visual vogler wang welch well were wiley with work ying zicheng
http://csdl.computer.org/comp/proceedings/cvpr/2004/2158/01/215810699abs.htm	93	Augmenting Images of Non-Rigid Scenes Using Point and Curve Correspondences Adrien Bartoli, Eugenie von Tunzelmann and Andrew Zisserman ´ Department of Engineering Science, University of Oxford {Bartoli,az}@robots.ox.ac.uk	active adjustment alexander algorithm algorithms analysis animation assuming bascle based best blake block blocks bombay bookstein brand bregler bundle cambridge chui clamping computed computer conference constant constraints convergence copenhagen correspondence cvpr decomposition deformable deformations denmark dense deriche derivatives described dynamic enforce equation equations estimation european evaluation exploit expression facial factorization features february figure flow frame from geodesic geometry good gool greece hartley hawaii hence ieee image incalibrated india initial intelligence international irani journal june kahl kanade kauai kerkyra koch linear machine manner mapping matching matrix method modelling models morphable motion multi multiple normal objects optical orthography pages paragios partitioning pattern plate point pollefeys pose press principal proceedings rangarajan rank realistic recognition reconstruction references region regions registration rigid scenes scharstein seattle separability september sequences shape shown simple since society solution solve space sparse spline splines stereo streams strong structure subspace surface szeliski taxonomy techniques that thin tikq time tomasi torresani tracing track tracking under understanding university unknowns using very video view vision visualization volume warps washington with workshop yang zisserman
http://csdl.computer.org/comp/proceedings/cvpr/2004/2158/01/215810187abs.htm	25	Simultaneous Calibration and Tracking with a Network of Non-Overlapping Sensors	activities adjustment alatas algebra algorithm algorithms analysis annual april artificial ascent automat automated ayache azarbayejani batch beymer bose building bulusu bundle calibration california camera cameras cannot castellanos causally cheeseman chiuso cholesky coifman common computer conf conference consistent convex coordinate correspondence corresponding cost counting csail cvpr definite dellaert department devices different distributed doherty each eccv editors element environment establishing estimating estimation estrin evaluation expo factor faugeras favaro fields fisher fitzgibbon focal frame framework from function ghaoui grimson ground gular hartley heidemann html http icmcs ieee ijcai infocom integrated intelligence international into introduces iterate jacobian javed joint july june kettnaker khan knight labeling langendoen least length less linear lncs localization machine maintaining malik many maps matrix mclauchlan measurement measuring media minka mobile modeling modern monitoring montiel motion moving multi multimedia multiple neira network networks newton next objects october optimize organization ostland outdoor over overlapping pages pami papers parameters partitioned pasula pattern pentland people performance philip pister placed planar plane position positioning positive practice probabilistic problem proceedings rabay rahimi randomly raphson rasheed real recognition reconstruction recover rectification recursive references relationships report representations requires ritov robot robotics robust romano russell savarese scene science sections seitz self sensor sensors september shah simultaneous small smith soatto society solving southern sparse spatial spmap springer squares statistics stauffer stein streams structure such surveillance synthesis system szeliski tards technical terms that theory thorpe three thrun tieu time tpminka tracked tracking traffic trajectories tran trans transactions trian triggs uncertain uncertainity university upper useful usenix verlag very video view vision visual volume wireless with without workshop zabih zisserman
http://csdl.computer.org/comp/proceedings/cvpr/2004/2158/01/215810429abs.htm	58	Hyperspectral Texture Classification Using Generalized Markov Fields Subhadip Sarkar and Glenn Healey Computer Vision Laboratory Department of Electrical Engineering and Computer Science University of California, Irvine Irvine, CA 92697	academic acknowledgments actually after ahuja also anal analysis analytical applicationof applications applied approaches arinen august avoid bagdazian baltimore bands based been besag better block boston buss campbell captured case chatterjee chellappa choice classification classify classifying cohen color comp comparing component compression comput computations computer conclusion conditional conf conference contains conventional cvpr cybern daily dallas data decision decreases desai detection diego discrimination does edition editor editors efficient energy environment estimation expect expected experimental experimentally extraction fact feature field fields filter filters first foundation from gaussian geometry golub good graphics grayscale gross have healey hopkins hyperspectral ieee illumination image imagery images imaging improve independent inform intell interaction interfere invariant jain john johns journal karhunen kashyap kluwer lack lattice loan machine magnitude manjunath markov matrix mean method methods misclassification model modeling modelling models moments more most moura multiscale multispectral national naval neighbors noise number observed obtain obtained office opponent optimal orlando other outperforms pages panjwani patt pattern peaking perform performance press principal proc proceedings process processes processing random rate recogn recognition references representation require research resemble results review royal rules samples schweizer science segmentation selection september showed significant significantly size sizes small society sons spatial spectroscopy speis spie standard statistical stochastic structure study suen suggested suggests supported symmetries synthesis syst systems tend texture textured textures thai than that theory thereafter these they third this trans tunable typical university unsupervised used using vector vetterli vision volume wang wavelet wiley with work would zernike
http://csdl.computer.org/comp/proceedings/cvpr/2004/2158/01/215810872abs.htm	116	A Unified Framework for Uncertainty Propagation in Automatic Shape Tracking	active adaboost aero akgul algorithm american anal analysis anandan annual appearance applied ashape association available banning barden based basedapproach behrenbruch bilinear blake boosting borders bosch british campo canada cardiac carne chichester clinical coarse comaniciu common component componentbased computational computer conf conference consider constraint constraints contour contours cootes coupledcontour covariance cristinacce cvpr czech dagm deformable descriptors detecting detection discriminant dublin echocardiographic echocardiography edition effect elect estimates estimation european experiments face facial factorization feature features fine flow framework france freund friedman function fused fusion geest generic german germany guided have heisele heteroscedastic http ieee image images imaging information intell intl irani ireland isard jacob john jones journal kambhamettu kanatani kanazawa kelion kendall krishnan krucinski learning leedan left lelieveldt linear local machine magdeburg manuscript matrices medical meer mikic mitchell model models motion nice noble noise object optical optimal optimization orthogonal otto papageorgiou paris part patt pattern patterns pedestrians philadelphia poggio pontil prague presence problems proc proceedings process projections quantifying really recog recognition references regional regression regularized reiber republic robust rules saunders schapire schmid segmentation sensor sequences serre shalom shape snow society sonka sons space spatial springer statistical syst system taylor textbook theories theory thomas through track tracking trackingmyocardial trainable trans ultrasound uncertainty unified unpublished using vancouver ventricular verlag viola vision vogelhuber volume wiau wiley with workshop zhou
http://csdl.computer.org/comp/proceedings/cvpr/2004/2158/01/215810151abs.htm	20	Camera Calibration from a Single Night Sky Image	aerospace analysis applications attitude autonomous axis based blank blanke calibration camera canon computer conference congress control cvpr determination dimage february figure finepix flexible fujifilm geometriccentroid html http ieee ifac intelligence june machine magnetic mathworld minolta pattern proceedings recognition references results rsted satellite sigma society system tamron technique three torquing transactions vision wisniewski with wolfram world zhang
http://csdl.computer.org/comp/proceedings/cvpr/2004/2158/01/215810120abs.htm	16	Space-Time Video Completion	analysis animation apllied approach approximate arya ballester based bertalmio bertozzi bobick caselles chan chiuso cohen comaniciu completion computer conference controlled criminisi curvature cuts cvpr data degraded dimensions doretto drori dynamic dynamics eccv efros elastica enhancement essa euler exemplar fast feature fitzgibbon fixed fluid fragment freeman from global graph graphcut graphics iccv ieee ijcv image inpaint inpainting irani joseph kang kokaram kwatra learning leung levin levoy lischinski math mcmc mean meer missing mixing motion mount movie navier nearest neighbor object occlusion pami parametric pattern peleg perez practical priors proceedings quantization queries quilting recognition references removal rendering resolution robust salesin sampling sapiro sapiroand scho schodl shen shift siam siggraph soatto society soda space sprites statistical statistics stokes synthesis szeliski texture textures toward toyama transfer transparency treatment treestructured turk using vcir vector video vision visualization weiss werman wexler yaniv yeshurun zisserman zomet
http://csdl.computer.org/comp/proceedings/cvpr/2004/2158/01/215810730abs.htm	97	Role of shape and kinematics in human movement analysis	accepted across activities activity advanced algorithm algorithms also analysis angles arma august automatic automatica avbpa baltimore based baseline between bissacco bobick body bone both bowyer build capability carter challenge chellappa chiuso chowdhury classification cock collins computations computer conf conference configuration cowdhury cues cunado cuntoor cutting cviu cvpr dalley data describes description determine different discriminating distance distances doretto drought dryden dynamic dynamical dynamics encode encodes enhance estimation events evidence example extraction face framework from future gait gaits gathering gavrilla gesture golub greater gross grother help hidden hoenkamp hopkins however human humans iccv icip icpr identification ieee image information integration interacting intersensory intl january john johns joint journal june kale kinematical kinematics kory krueger labelling learning like loan mardia markov math matrix methods model models moor more motion movement murray nash networks nixon normal objects overschee pages parameters pattern patterns pedestrian perceive perception perceptual perform performance person phillips poorer possibly power press problem proc proceedings processing proffitt rajagopalan recognition references refinement relation require robledo sarkar sensory sequences sets shape significantly silhoutte simple soatto society sons speeds statistical stochastic studies study subspace such sundaresan surgery survey symp system systems tanawongsuwan task technique textures than that theory tieu time tolliver transactions under university using vaswani vision visual walking wiley wish with
http://csdl.computer.org/comp/proceedings/cvpr/2004/2158/01/215810082abs.htm	11	Bayesian Assembly of 3D Axially Symmetric Shapes from Fragments	absolute accommodate accurately acva additional alignment america anal analysis annals appear application approach appropriate arbitrary archaeology assemble assembled assembling assembly automatic available axially based bayesian because been belongs bern better between bolle british broken brown chapter classification closed column combining comp completely complex comput computational computationally computer conceptual conceptually conclusion conf conference continuous cooper correctly cultural curve cvpr data december deformation difficult dimensional discussed doing each easier electron estimates estimating estimation faster fewer field figure fitzgibbon form found fragmented fragments frames framework free freeman from garder geometric geometries geometry global goldberg graph group grouping having hawaii hence heritage horn http icpr ieee ijcv image importance important information instead intell international into jigsaw journal june kalvin karabassi kimia know known kong lambdan large learning least leit less machine make malon markov matched matching mathematical measurements method methods modeling models more much multiscale necessary noisy numbers object objects only operations optical optimally option orientation outlined pages pami papaioannou paper parameters partial parts pattern patterns pennec perceptual pictorial pieces point points position possible pots present presented priori problem problems proc proceedings prohibitive puter puzzles quaternions random range rather reached reality reassembly recognition reconstruction references registration representations research result robust schonberg sculpture sets shape shapegrabber shapes sherd sherds should shown since sites small society solution solving springer squares stage stochastic stolfi such surface symmetric symposium that their then theoharis thesis thirion this three through toroslu trans transformation treat ucoluk umeyama uncertainty unit university unknown unsupervised using validation vast very view virtual vision volume were where which willis with wolfson work workshop would
http://csdl.computer.org/comp/proceedings/cvpr/2004/2158/01/215810528abs.htm	70	Detection and Removal of Rain from Videos	across address algorithm algorithms along also analysis angled appearance applications atmosphere atmospheric august averaged back background based beard because been between book bound bounds bright cambridge camera case causes challenging change changes chiuso chose chuang clarity class clifford clip clouds columbia complex complexities comprehensive compute computed computer conclusion conditions conducted conference conservative consider consists constant contrast contribution correlation courtesy creases currently cvpr defocused degrade demonstrate dependence derained derive despite detect detected detection detects developed difference differences different direction distance distribution does doretto drop droplet dynamic editing effected effective effectiveness effects efficient electromagnetic entertainment equation equilibrium even example examples experiments explicitly exposure expression extend factor fall falling fast figure filtering focal folds forming forth frames frequencies from further future fuzzy garg given glass graphics gunn hail handle have head hence hill home horn however iccv ieee ijcv illustration image images implications including indexing induced insensitive intensities invariant issues john josa journal kept kinzer larger length line magnolia make manning maps marshall mason maximum mccartney mcgraw measure meterology method methods model modeling models molecules motion motions moves movie moving narasimhan nayar needle neglected negligible neighborhoods note object objects obtain obtained only optical optics original other over palmer pans parameters particle particles passing path pattern person phone photometric physical pixel pixels please pool possible presented preserving press proceedings produced productions projects properties propogation proposed purposes quality rain raindrops rainfall rainmaking range ranges realistic recognition reeves references region regions related removal remove removes removing rendering report represented represents results retrieval right ripples robot robustly same scaled scattering scene scenes schematic science scintillations screen segmentation seven several severely shape shirt shown shows siggraph similar simple simulation since size sizes snow soatto society sons sparse spatially spatio speaking stagnant starik stays steady stochastic streak streaks strength substituting subtle such surveillance system task techinque technical temporal terminal terms texture textures than that then these they this those three threshold through thus time triangle types typically unable university upper used using value values varying velocity very video videos viewed viewing visible vision visual wall wang water weak weather were werman when where white wide width wiley window wish with within without words work workshop zoom zooms
http://csdl.computer.org/comp/proceedings/cvpr/2004/2158/01/215810342abs.htm	46	The SPS Algorithm: Patching Figural Continuity and Transparency by Split-Patch Search	about accurate achieved actual adapted advantage algorithm algorithms aliasing analogies analysis appear applying approach approximately artefact artefacts available background based bayesian been birchfield blake bornard both boundaries boundary broadhurst cambridge carving case central check chenot chuang cipolla combines comp complex compositing composition computational computer computing conclusion conf conference consistency constrain continuity contribution copenhagen correction correlation correspondence criminisi cstr curless cuts cviu cvpr cyclopean czech data december dense depth desired detects digital disparity dissimilarity down dual eccv effects efficiency efficient efros emphasis employ enhances estimation eugene europe evaluation examples exemplar exemplarbased exploits faithful fashion february figural figure first fitzgibbon fiume fixed foreground frame france free from gaze generated geometry golland graph graphics harrison have head here hertzman hierarchical hingorani hirschmueller iccv ieee ijcv image images implementation improvements inpainting input insensitive instructions intelligence jacobs koch kolmogorov kutulakos laborelli lack layers lecan lecture left leung lncs looking machine madison maggs main make manipulation matching matting maximumlikelihood means measure microsoft missing mixed mixing more moreover multimedia multiple nice notes novel numerous object objects occlusions oliver only pages paper parametric pass patch patchbased patches pattern pentium perez performed phase pixel pixels plzen presented priors proc procedure proceedings processor produces quality real realistic recognition reconstruction reduces references refinement removal removes rendered rendering replacing report republic research result results right rochester runs salesin sampling scharstein science search seitz sequences shape sharp shotton show shown siggraph simple society space speed split springer static statistical step stereo stereoscopic still stuck substitution surface synthesis synthetic szeliski taking taxonomy technical technique techniques teleconferencing temporal texture that them theory this thresholding thus tightly time tomasi torr toyama transactions transitions transparency university unnaturally upon using verlag view views virtual vision visual volume wexler which with yielding zabih zisserman
http://csdl.computer.org/comp/proceedings/cvpr/2004/2158/01/215810379abs.htm	51	Robust Color Object Detection using Spatial-Color Joint Probability Functions David Crandall Jiebo Luo	acknowledgements active addison additional addresses aggarwal alarms algorithm allow alternative ambiguity analysis appealing appearance applied area assign assigned authors ballard baluja based beneficial best body build bureau capable cars cause cech cechs challenging chang choice class color colors combining comparing comparison compound computer conclusions conf conference containing cooccurrence cootes coping correlograms could crafting curve cvpr dataset demonstrated detecting detection dictionary differences different difficult digital discerning distance distortion edge edwards efficacy either encoding european evaluation example experimenters experiments extremely face faces false feature fergus figure fleck forms forsyth fragile froc from functions future fuzzy geoff geometric giorgianni good government gross handling hausdorff help histograms huang human hunter huttenlocher ieee image images improvement incorporate indexing information instances intelligence international intl into invariant iqbal john joint journal judd kanade kelly klanderman kodak krumm kumar language layout learning light like lines machine madden management mechanisms method mitra model models most motivated multiple names naming national negligible nelson network neural number object objects occurrence offering office often only operating opportunities oren original osuna over overhead papageorgiou paper pattern pedestrian perceptually performance permission perona pink pixel plans poggio point positives possible printing probability problem proc proceedings produces proposed provide publication quantization quantize quantized reading recognition records references represents results retrieval richer ricklidge robustness rotation rowley rule same scale schneiderman selinger several shape significant single sinha society solution solutions some sons space spatial specific standards statistical statistics still structure study subtle swain takes taylor templates text texture thank that their there this transactions type unconstrained universal unsupervised using variations vary views vision washington wavelet wesley while white wiley wish with woolfe work york zabih zisserman
http://csdl.computer.org/comp/proceedings/cvpr/2004/2158/01/215810016abs.htm	2	Wide Baseline Feature Matching Using the Cross­Epipolar Ordering Constraint Xiaoye Lu and Roberto Manduchi Department of Computer Engineering, University of California, Santa Cruz1	across actual affine algorithm algorithms angle applications artificial baker based baseline bauckhage baukhage baumberg binford bipartite blue bombay buhmann cambridge candidate chain clifford coarse color commission computer computing conditions conf conference cormen corner correspondence corresponding costeira critical cvpr databases dellaert depth deriche descriptors detectors dimensional direct dissimilarity distributions dynamic eccv edge empirical epipolar evaluation faugeras feature features ferrari figure fine flipping from geometry global gool graphs grayvalue graz greece guibas hartley hierarchical hopcroft iccv ieee ijcv image images implementation india insets intelligence intensity inter interest international intra introduction invariant invariants isprs itai january joint journal kanade karp kerkyra learning leiserson line lines local lowe luong machine maciel madison matching maximum mcmc measures meerbergen metric mikolajczyk modeling mohr moons motion multi multiple object ohta optimal pages pami part pattern patterns performance photometric planar point points pollefeys press pritchett problem proc proceedings profile programming puzicha recognition recovery references registration reliable results retrieval right rivest robust rodeh rubner scale scanline schmid schmidt search seconds seitz separated show siam society solution some steero stein stereo stockhom strecha structure sweden table tanimoto texture thorpe three thrun time times tomasi towards trans tuytelaars uncalibrated under ungureanu university unknown using vergauwen view views vision wide widely wisconsin with workshop wyngaerd zhang zisserman
http://csdl.computer.org/comp/proceedings/cvpr/2004/2158/01/215810275abs.htm	37	Super-Resolution Through Neighbor Embedding Hong Chang, Dit-Yan Yeung, Yimin Xiong Department of Computer Science Hong Kong University of Science and Technology Clear Water Bay, Kowloon, Hong Kong {hongch, dyyeung, csxym@cs.ust.hk}	acoustics advances allebach ames analogies analysis angeles annual applications approach april arbitrary artificial atkins august baker ballester based bayesian becker belkin bertalmio between bird bishop bouman branch break cally cambridge capture carmichael caselles center cheeseman clustering computation computer conference convolution cubic curless cvpr data december definition dietterich digital dimensional dimensionality dynamic editors efros eigenmaps eighth elad embedding error examplebased expansion faces fast feuer field figure framework freeman from function generated geometric ghahramani global globally graphics greece ground hallucinating hallucination hanson head hertzmann hilton ieee image images improved information inpainting intelligence interactive international interpolation island jacobs jones jordan journal july june kanade kanefsky kauai kearns kerkyra keys kraft langford laplacian learning leung level levoy limits linear lizard local locally machine madison manifold manifolds mapping model moffett multiple nasa nearest neighbors neural niyogi nonlinear nonparametric number obermayer observations oliver orleans over pages parametric pasztor pattern perceptual plant press primal priors proceedings processing quality quantization recognition reconstruction reduction references report representation research resolution resolved roweis salesin sampling sapiro saul savannah schultz science september sequences seventh shum siggraph signal silva sketch society solla spectral speech step stevenson storkey structure structured stutz super superresolution surface surfaces synthesis systems technical techniques tenenbaum texture them think thrun tipping transactions tree truth twenty unsupervised used using vector vision volume with zhang zheng
http://csdl.computer.org/comp/proceedings/cvpr/2004/2158/01/215810776abs.htm	103	Multibody Motion Segmentation Based on Simulated Annealing	analysis asian baltimore based body component computations computer conference consistency costeira criterion cvpr decomposition discriminant dynamic eccv edition england face factorization feature figure from gear generalized golub good gpca grouping groups hopkins huang ichimura ieee image images independent international irani john kanade kanatani loan machline manor matrix method methods model modeling motion moving multi multibody objects orthogonal orthography pattern press principal proc proceedings recognition references research respectively revisiting sastry scenes segmentation selection separation sequence shape shown society streams studies subspace three tomasi track under university vidal vision with workshop zelnik zhang zhou
http://csdl.computer.org/comp/proceedings/cvpr/2004/2158/01/215810212abs.htm	28	Capturing image structure with probabilistic index maps	able academic accuracy algorithm algorithms aligned allows also appearance applied automatically background based been benefits better beyond both btij bumps captures cases change changes channel choice classes clear clsutering cluster clustering clusters collection color column compare compares comparison completely complex complexity computation computer concatenates conditions conf conference continuous contrast cost covariance cvpr data dataset decomposition depends detection different distributions drastic each easier energy entire entries entry equivalent error estimated even examples existing expected experiments extended extra faces fashion favorably feature features figure find flexible fore foreground form frame free frey from gaussians ghahramani given graphical gray grimson ground grouped handle happening here hood however identities ieee illumination illustrate image images include increased incremental index inference inferred information input instantaneous intensities intensity interesting into introduction invaraint invariance invariant jaakkola jojic jordan kluwer last layers leads learned learning level local makes malfunction many maps match means measurements methods middle miller minimizing mixture mixtures model models more much multiple neural nips nonoverlap norwell note number occurs only other palette pami paper pattern people ping pixel poorer posterior probabilistic probability proc proceedings processing profile published publishers rate rates rather recognition recomputed recovery references regular remotely report representation represented representing respectively result resulting results rows saturation saul section selection separate separating show shown shows similar similarity single situations small society sprites standard start stauffer stij stream structures subset subtraction successful such superior switch synthesized systems tasks technique techniques templates terms test than that then this three tieu tmpim tracking trained training trans transformation transformed transforminvariant typical under unsupervised used using variability variance variational various vector video vision were what wise with would
http://csdl.computer.org/comp/proceedings/cvpr/2004/2158/01/215810552abs.htm	73	Wide-baseline Stereo from Multiple Views: a Probabilistic Account Christoph Strecha, Rik Fransens, Luc Van Gool ESAT-PSI, University of Leuven, Belgium {christoph.strecha,rik.fransens,luc.vangool}@esat.kuleuven.ac.be	affinely algorithm algorithms alvarez approach based baseline bayesian bmvc cambridge carving chum complete computer conference correspondence cvpr data dempster dense deriche derivatives diffusion eccv efficient embased empirical estimation evaluation external faugeras filtering fitzgibbon frame from geometry gool haar hans hartley hulls iccv ieee ijcv image incomplete invariant jvcir keriven kutulakos laird level likelihood lippman local matas matching maximally maximum methods motion multiple nonlinear pajdla parity pattern photo press priors proc proceedings processing recognition references regions reliable rendering respecting robust romeny rubin sanchez scale schafer scharstein schemes segmentation seitz shape slabaugh society space stable statist stereo stereovision strecha symp szeliski taxonomy theory transactions transmission tuytelaars university urban using vasconceles viergever view views vision visualization weickert wexler wide zisserman
http://csdl.computer.org/comp/proceedings/cvpr/2004/2158/01/215810372abs.htm	50	Hybrid Textons: Modeling Surfaces with Reflectance and Geometry	analysis another appear appearance august bidirectional both bottom center chantler compact compare comparison computer conference contain crumpled cula cvpr dana december direction distribution dong doorn each example feature figure five frontal function functions geometric ginneken goal graphics ground histogram histograms hybrid ieee image images integration international january journal june koenderink left method methods model modeling nayar object pages paper pattern pitted plaster predict predicted proceedings real recognition references reflectance reflection representa result right rightmost rough rows sampled society surface surfaces synthesis texton texture textured textures these this thoroughly tion transactions truth using viewing vision with workshop world
http://csdl.computer.org/comp/proceedings/cvpr/2004/2158/01/215810106abs.htm	14	Region-Based Progressive Stereo Matching Yichen WEI Long QUAN Department of Computer Science, Hong Kong University of Science and Technology Clear Water Bay, Kowloon, Hong Kong {yichenw,quan}@cs.ust.hk	about accuracy acknowledgements actually after algo algorithm algorithms almost although ambiguity among analysis application approach approximate april asian based bear belief believe between binary birchfield bleyer bobick boundaries boykov camera canada caused chen collins colorado column comaniciu comparison component computation computer conclusions cone cones conf conference consistency consistently constraints cooperation copenhagen correspondence cuts cvpr data decision denmark dense density depth details developed disc disparity dissimilarity dynamic each early easily efficient energy environments ernst error estimation european evaluation exact except experiments extracting fast fastest feature features figure final finally finding finely flexible flowers fort found frame framework france from gcmulcam gelautz glob global gong good grant graph greece greig ground handled high hkust hong http icip ieee image imagebased images including incomplete initial initially insensitive integrates intelligence international intille into iteration iterations jeju journal july kambhamettu kerkyra kolmogorov kong korea large largest layered left level lhuillier light limited local machine mainly maps match matching maximum mean measure medioni meer method methods middle middlebury minimization model modeling most motion multi multiple multiway nice novel number object observed obtained obvious occlusion occlusions only order other overveld pages paper parameter pattern perform performance performs pixel plotted porteous possible posteriori presented proceedings produced programming progressive progressively prop propagation propogation quan quasi rank ranked ranking rate reasonable recognition reconstruction recovered reference references region regions relative relatively reliability reliable rendering representation result results rithms robust royal running runs sampling sara satisfactory sawhney sawtooth scale scene scharstein scheme second seconds segm segment segmentation seheult semi september series sets setting shan shift should shows shum since size slanted smile society space state statistical stereo sticks structure structured structurefrom submitted subscript supported surface surfaces symmetric szeliski table taxonomy teddy terms tested textured that their them thin thirty this those time tomasi toward trans truth tsukuba unambiguous untex using vancouver veksler venus versus very visibility vision visual visually volume volumetric well which wilinski with work workshop wrong yang zabih zhang zheng
http://csdl.computer.org/comp/proceedings/cvpr/2004/2158/01/215810692abs.htm	92	High Resolution Video Mosaicing with Global Alignment Roberto Marzotto, Andrea Fusiello, Vittorio Murino Dipartimento di Informatica, Universita` degli Studi di Verona 37134 Verona, Italy {andrea.fusiello,vittorio.murino}@univr.it	accuracy acknowledgments adjustment algorithm alignment also analysis application applications approach april artificial august auto automated automatic automation available barcellona based better bounds bundle calibration camera capel carnegie chio cohen colombari compared computation computer conclusions conference confirm construction contributed contributions cost courtesy cvpr davis defence described detection different discussions energy engineering environments estimation europe european existing experiments features ference fewer figure flexible focusing frame from function functions fusiello global goodness graph graphical graphics greece hartley high homography ieee image implementation improving include inference intelligence international introduced irani issues iteration iterations iterative joint journal kanade kanatani kang kumar laplacian layered left local lucas machine marcelo march measure medioni mellon method methods models mosaic mosaicing mosaics moving naoya nathaniel nikzad novel number objects odone ohta optical optimal pages panoramic pattern peleg pittsburg pixel planar point practical presented proceedings processing projective provided quality recognition references registration report representation resolution right robotics robust sawhney scenes sept september sequences shimizu shot shum smolic society solutions spain starting stereo subbarao super szeliski technical technique techniques thessaloniki this three through tomasi topology tracker tracking transactions transformations tree trucco university useful using value video virtual vision volume which wiegand wise with work yields yoshiyuki zeno zhang zisserman zoom