成像目標跟蹤數(shù)據(jù)關聯(lián)技術研究.pdf

1、東南大學碩士學位論文成像目標跟蹤數(shù)據(jù)關聯(lián)技術研究姓名：赫沙申請學位級別：碩士專業(yè)：控制科學與工程；模式識別與智能系統(tǒng)指導教師：金立左20120329ABSTRACTThetechniqueofdataassociationhasplayedallimportantroleintheresearchofvideotargettrackingItisusedtoassociateandcombinemultiplesourcedataand

2、itcangainmoreaccuratedataSOastoimprovethetargettrackingperformanceThefocusofthisdissertationistheapplicationofdataassociationinvideotargettrackingAimattheproblemsofsingletargetandmulti—targettrackinginvideosequence，thete

3、chniqueofdataassociationisimplementedandimprovedbyusingbasictheoriesandmethodsInthisdissertation，videotrackingsystemanditsbasicprinciplesarebrieflyintroduced，andsomeexperimentsofclassicdetectionmethodsareperformedBaseont

4、heratherdetailedanalysisofdataassociation，theissuesoftrackinitiation，trackmaintenance，trackterminationandtrackingdoordesigninghasbeenexplainedThedissertationhasmadeafurtherresearchonthealgorithmofNN，PDA，PFGNN，MHTetcBaseo

5、npositioninformationofimagetargetusingindataassociation，thetrackingissteadyandtrackingprecisionisimprovedInsingletargettrackingbyusingPDA，AdetailedanalysisoninfluencefactorstoperformanceisgiveninthedissertationThequantit

6、yandstateoftargetsinmulti—targettrackingareunknownandfrequentlychangingBasedontherandomfinitesetsstatistics，theproblemissolvedbyprobabilityhypothesisdensity(PHD)filterPHDfilterusingSequentialMonteCarloanditsrealizationpr

7、ocessisgiveninthisdissertationPHDCanmakefalseassociationbecauseitcannotgettheassociationinformationofalltargets，thus，ithascombinedwithGNNThesimulationresultsshowthatthetechniqueofdataassociationhasminimizedoccurrenceoffa