SMC-PHD based multi-target track-before-detect with nonstandard point observations model.pdfVIP

J．Cent．SouthUniv．(2015)22：232—240 垒Springer D0I：10．1007／sl177l-015．2514．x SMC—-PHDbasedmulti--targettrack-before--detectwith nonstandardpointobservationsmodel ZHANRong．hui(~荣辉)，GAPYon．zhao(高彦钊)，HUJie．min(胡杰民)，ZHANGJun(张军) ScienceandTechnologyonAutomaticTargetRecognitionLaboratory， NationalUniversi够ofDefenseTechnology，Changsha410073，China ◎CentralSouthUniversityPressandSpringer-VerlagBerlinHeidelberg2015 Abstract：Detection andtrackingofmulti-targetwithunknownandvaryingnumberisachallengingissue，especiallyunderthe conditionoflowsignal—to．noiseratio(SNR)．Amodifiedmulti．targettrack—before—detect(TBD、methodwasproposedtotacklethis issueusinganonstandardpointobservationmodel_Themethodwasdeveloped from sequentia1MonteCarlo (SMC)一based probabilityhypothesisdensiyt (PHD)filter,anditwasimplementedbymodifyingtheoriginalcalculationinupdateweightsofthe particlesandbyadoptinganadaptiveparticlesamplingstrategy．ToefficientlyexecutetheSMC—PHD basedTBD method，afast implementationapproachwasalsopresentedbypartitioningtheparticlesintomultiplesubsetsaccordingtotheirpositioncoordinates in2D resolutioncellsofthesensor．Simulationresultsshow theeffectivenessoftheproposedmethodf0rtime—varyingmulti—target trackingusingraw observationdata． Keywords：adaptiveparticlesampling；multitargettrack—before—detect；probabiliythypothesisdensiyt (PHD)filter；sequential MonteCarlo(SMC)method modelsandmoresuitableofrrea1．timeapplication．0ne 1Introduction oftheeffectivewaystoimplementBayesianrecursionis thesequentialMonteCarlo 1『5—161(SMC，alsoknown Intraditionaltargettracking．adetectionthresholdis a