预览加载中，请您耐心等待几秒...

Mean-shift跟踪算法及其在光电跟踪系统中的应用.docx

预览
1/2
2/2

在线预览结束，喜欢就下载吧，查找使用更方便

下载提示 文本预览

如果您无法下载资料，请参考说明：

1、部分资料下载需要金币，请确保您的账户上有足够的金币

2、已购买过的文档，再次下载不重复扣费

3、资料包下载后请先用软件解压，在使用对应软件打开

Mean-shift跟踪算法及其在光电跟踪系统中的应用 Introduction: Mean-shifttrackingalgorithmisasimpleyetpowerfultrackingtechniquethathasgainedtremendouspopularityinrecentyearsacrossvariousfieldslikecomputervision,objectrecognition,imageprocessing,andsurveillancesystems,etc.Thisalgorithmiswidelyusedinoptical-electronictrackingsystems,especiallyintargettrackinginvideosequences. Mean-ShiftTrackingAlgorithm: Mean-shiftisaniterativealgorithmthatfacilitatestrackingofthetargetbyupdatingthecenterofgravityoftheprobabilitydistributionfortheobject’sappearance.Inotherwords,themeanshiftalgorithmcalculatesthedisplacementvectorthatisrequiredtoshiftthecentroidofthetargettothecurrentlocation.Themean-shiftalgorithmperiodicallyupdatesthetarget’spositionandadjuststhewindowaroundthetargetaccordingly. Themean-shiftalgorithmismainlyappliedintwosteps:first,“targetmodel”iscreated,andthen,“searchwindow”isdefined.Inthefirststep,thetargetmodelisdefinedasahistogramofcolor,texture,orfeatures.Inthesecondstep,thesearchwindowisdefinedaroundthetargetareainthepreviousframetodetectthetargetregioninthecurrentframe.Mean-shiftcalculatestheprobabilitydistributionfortheobject’sappearanceaccordingtothetargetmodelanddefinesthesearchwindow.Iteratively,thealgorithmtriestoconvergetothecenteroftheprobabilitydistributionbyupdatingthecurrentpositionofthetargetuntilconvergence. ApplicationofMean-ShiftTrackinginOptical-ElectronicTrackingSystem: Mean-shifttrackingalgorithmhasbeenwidelyusedinOptical-Electronictrackingsystems,especiallyinthetargettrackingofvideosequences.Inoptical-electronictrackingsystems,thecameracapturestheimageofthetarget,andtheimageisprocessedbythesystemtotrackthetarget.TheMean-Shiftalgorithmisusedintheimageprocessingcomponentofthesystemtotrackthetargetobjectframebyframe. ThesuccessoftheMean-Shiftalgorithminoptical-electronictrackingsystemscanbeattributedtoitsspeedandrobustnessinreal-timetrackingoftargets,evenwhenthereisachangeinthebackground,lighting,orscaleofthetarget.TheMean-Shiftalgorithmishighlyefficientintrackingb