A Framework for Human Tracking using Kalman Filter and Fast Mean Shift Algorithms.pdf

《A Framework for Human Tracking using Kalman Filter and Fast Mean Shift Algorithms.pdf》由会员上传分享，免费在线阅读，更多相关内容在学术论文-天天文库。

1、AFrameworkforHumanTrackingusingKalmanFilterandFastMeanShiftAlgorithmsA.AliK.TeradaGraduateSchoolofAdvancedTechnologyandGraduateSchoolofAdvancedTechnologyandScience,UniversityofTokushima.Science,UniversityofTokushima.Tokushima,JAPAN.Tokushima,JAPAN.ali@is.t

2、okushima-u.ac.jpterada@is.tokushima-u.ac.jpAbstractalgorithmsdetectschangedregionsbymeansofedgecomparisons[3].ThetaskofreliabledetectionandtrackingofmultipleTheKalmanfilterhasbeenextensivelyusedintheobjectsbecomeshighlycomplexforcrowdedscenarios.Invisionco

3、mmunityfortracking.BroidaandChellappa[5]thispaper,arobustframeworkispresentedforusedtheKalmanfiltertotrackpointsinnoisyimages.Inmulti-Humantracking.Itincludesacombinationofstereocamera-basedobjecttracking,BeymerandKonoligeKalmanfilterandfastmeanshiftalgori

4、thm.Kalman[6]usetheKalmanfilterforpredictingtheobject’spositionpredictionismeasurementfollower.Itmaybemisledbyandspeedinx-zdimensions.RosalesandSclaroff[7]usewrongmeasurement.ThesearchforsolutionisguidedbyatheextendedKalmanfiltertoestimate3Dtrajectoryofanf

5、astmeanshiftprocedure.Itisusedtolocatedensitiesobjectfrom2Dmotion.Acommonapproachtohandleextrema,whichgivescluethatwhetherKalmanpredictioncompleteocclusionduringtrackingistomodeltheobjectisrightoritismisledbywrongmeasurement.Trackingmotionbylineardynamicmo

6、delsorbynonlineardynamicsresultsaredemonstratedforcrowdedscenesandand,inthecaseofocclusion,tokeeponpredictingtheevaluationoftheproposedtrackingframeworkisobjectlocationuntiltheobjectreappears.Forexample,apresented.linearvelocitymodelisusedinBeymerandKonoli

7、ge[6]andaKalmanfilterisusedforestimatingthelocationandmotionofobjects.Fortheimagesegmentationproblem,1.IntroductionMean-ShiftClusteringiscommonlyused.ComaniciuandMeer[8]proposethemean-shiftapproachtofindclustersMultipleobjecttrackingisanextensivelyinvestig

8、atedinthejointspatialandcolorspace.Arecentarticlebysubjectinthefieldofvisualsurveillance[1].ItsmainPulford[9]summarisesthetechniquesinwidespreadusecomplexitystemsfromthefactthatobserveddataisusuallyandclassif