It is a difficult task to find an observation model that will perform well for long-term visual tracking. In this paper, we propose an adaptive observation enhancement technique based on likelihood images which are derived from multiple visual features. Th
ROBUSTOBSERVATIONSFOROBJECTTRACKING
BohyungHanandLarryDavisDept.ofComputerScienceUniversityofMarylandCollegePark,MD20742,USA{bhhan,lsd}@cs.umd.edu
ABSTRACT
Itisadif culttaskto ndanobservationmodelthatwillperformwellforlong-termvisualtracking.Inthispaper,weproposeanadaptiveobservationenhancementtechniquebasedonlikelihoodimageswhicharederivedfrommultiplevisualfeatures.ThemostdiscriminativelikelihoodimageisextractedbyPrincipalComponentAnalysis(PCA)andincrementallyupdatedframebyframetoreducetemporaltrackingerror.Intheparticle lterframework,thefeasibil-ityofeachsampleiscomputedusingthismostdiscrimina-tivelikelihoodimagebeforetheobservationprocess.Inte-gralimageisemployedforef cientcomputationofthefea-sibilityofeachsample.Weillustratehowourenhancementtechniquecontributestomorerobustobservationsthroughdemonstrations.
1.INTRODUCTION
Trackersarebasedonthesomemeasurementofsimilaritybetweenthetargettobetrackedandobservations,andvar-iousobservationmethodsareusedtode nethissimilarity.Intensityorcolorisnaturaltouseinobjecttracking,andap-proachesbasedontemplates[1]andhistograms[2,3,4,5]areverycommon.Also,edges[8]and lterresponses[9,6]areimportantfeaturesforobjecttracking.Variousobserva-tionstrategieshavebeenproposed,butthereisnogenerallysuperiorobservationmethodforgeneralvisualtrackingal-gorithms.So,weinsteadproposeanobservationenhance-menttechniquebasedonlikelihoodimages,whichcanbeincorporatedintomanytrackingalgorithms.
Alikelihoodimagerepresentsthecontrastinformationbetweenforeground(targetregion)andbackground(itssur-rounding);itiscreatedbycomparinghistogramsofbothareaswithrespecttosomefeatures.Itwasoriginallysug-gestedin[10]fortrackingproblems;there,themostdis-criminativefeatureselectedfromasetoflikelihoodimagesisdirectlyusedformean-shifttracking.However,thisap-proachmayexhibitpoorperformanceinclutterandcanlosethetargetinspiteofitsvisualsalience.Also,thelikeli-hoodimagecanbesigni cantlycontaminatedbytemporary
trackingerrors.Therehavebeensomecloselyrelatedworks[7,6],buttheydonotprovideanadequatesolutiontotheseproblems.
Inthispaper,wepresentanobservationenhancementtechniqueusinglikelihoodimagesobtainedfromtwodif-ferentfeaturespaces–RGBandnormalizedRGB(rgb).Sixlikelihoodimagesarecreated,andamostdiscrimina-tivelikelihoodimageisextractedbyPCA.Inordertoavoidpollutionoftheextractedlikelihoodimagebytrackinger-ror,weupdatethesubspaceincrementallyontheassump-tionthatthescenearoundthetargetchangessmoothly.The“brightness”inthemostdiscriminativelikelihoodimagede-liverspriorinformationaboutthetargetregion,anditisem-ployedtocomputethefeasibilityofasample(tobede nedrigorouslybelow)beforethemeasurementstepinthepar-ticle lter.Thefeasibilityisobtainedbythesummationofvaluesinsidetheregioninthemostdiscriminativelikeli-hoodimage,andanintegralimage[12]isemployedforef- cientcomputation.The nalweightofeachparticleisde-terminedbytheproductofthefeasibilityandthelikelihoodinobservation.Asaresult,severalindependentfeaturesaremergedthroughthelikelihoodimagesandthemergedfea-turesareutilizedforrobustobservationviathefeasibilitycomputation.
Thispaperisorganizedasfollows.Wedescribethelike-lihoodimagesandthefeatureextractioninsection2and3,respectively.Insection4,trackinginaparticle lterframe-workandexperimentalresultsaredemonstrated.
2.LIKELIHOODIMAGES
Likelihoodimagesrepresentthedistinctivenessofatargetobjectfrombackgroundwithrespecttoagivenfeatureorsetoffeatures.Fortheconstructionoflikelihoodimages,log-likelihoodratiosareobtained rstfromhistogramsofforegroundandbackgroundpixels.Then,thesalientregioninforegroundcanbedetectedbyidentifyinghighlikelihoodratios.
Indetail,supposetheforegroundisgivenandtheback-groundisregardedasarectangularregionsurroundingthe
ROBUST OBSERVATIONS F... 暂无评价 4页 免费 Robust Object Tracking w... ...In their approach, translational models are used for tracking, which provides...
Video object tracking ba... 暂无评价 6页 免费 ROBUST OBSERVATIONS F... ...method for target tracking of multiple points on an object by using ...
Object tracking can be formulated as a continuous...previous observations and estimated state transitions...Indeed, learning a robust observation model has ...
We use parametric shape to model object contours. Observations are collected ...dynamic environments, which is very important for a robust tracking system. ...
Robust Object Tracking with Online Multiple Instance Learning Boris Babenko, Ming-Hsuan Yang, Serge Belongie. Robust Object Tracking with Online Multiple ...
Affine Object Tracking w... 暂无评价 12页 1财富值 Robust Object Tracking ...Abstract. A basic requirement for a practical tracking system is to adjust ...
It is the bifactor that makes the new robust estimator for correlated observations di?er from that of the IGGIII scheme or that of the LS method. 3 ...
tightly integrates object detection and tracking to guarantee a robust and e...the largest posterior probability based on current and previous observations. ...
Multiple Object Tracking Using Particle Filters M....observations for tracking the statistical distance ...Russell, “Toward robust automatic traf?c scene ...
May 21-26, 2001 Robust Object Tracking Using an Adaptive Color Model Gi-...Face tracking is used not only for directing the computer’s attention to ...
热门文档