基于粒子濾波的機(jī)器人視覺跟蹤研究與實(shí)現(xiàn).pdf

1、大連理工大學(xué)碩士學(xué)位論文基于粒子濾波的機(jī)器人視覺跟蹤研究與實(shí)現(xiàn)姓名：劉陽申請(qǐng)學(xué)位級(jí)別：碩士專業(yè)：控制理論與控制工程指導(dǎo)教師：莊嚴(yán)20071216大連理工大學(xué)碩士學(xué)位論文ResearchandImplementationofRobotVisualTrackingbasedonPanicleFiltersAbstractAsachallengingresearchtopicinthefieldofroboticsvisualtracking

2、iswidelyappliedtomilitary，visualsurveillanceanddigitalmedicaltreatmentBasedonreeursiveBayestheorycoloredtargettrackingformobilerobotwithmonocularvisionisdiscussedinthispaperAcomprehensiveresearchworkiscarriedoutinobjectd

3、etectionandobjecttrackingwhicharecentraltechniquesinmobilerobot’svisualtrackingindifferentenvironmentsAccordingtothesetrackingalgorithmswithdifferenttargetmodelsmobilerobotcanaccomplishthetrackingtasksinbothstationaryand

4、movingcasesAgroupoftargetdetectionmethodsareintroducedandanalyzedonthebasisofcolorvisionandcolorimetrytheoryConsideringsystem’srequirementinreal—timeprocessingandpracticalapplication，twodetectionalgorithmsbasedonweighted

5、colorhistogramandInnlengthencodearestudiedTherepresentationofBayesiantrackingisgivenbasedonMarkovhypothesisandBayesianformula，andtwoofitscommonimplementationmethodsareintroducedinsuccessionAimingatthetrackingtaskinnon—Ga

6、ussianmeasurementenvironment，MonteCarloMethodsandparticlefilterareintroducedindetailInordertoimplementtherealtimetrackingonthemobilerobot，anovelapproachofhybridparticlefilteralgorithmispresentedtoprocessthetarget’spositi

7、onandshaperespectively，whosestatesupdatingisonthebasisofdatafusionbetweenKalmanfilterandparticlefilterTheproposedmethodcannotonlypavethewayforalow—complexityparticlefilteralgorithmindealingwithhigherdimensionaltrackingpr

8、oblem，butalsocovertheshortageofGaussianrestrictioninKalmanfilterToensuretheeffectivenessofobjecttrackingwhentherobotisnavigating，anobjectpredictionmethodisproposedteducetheimpactcausedbythemovementofthecamerawhichisdeduc

9、edfromthecamera’simagingprocessDuetothevariationsofthetargets’features，differenttargetmodels，detectionmethodsandtrackingalgorithmsareemployedtoachievesuccessfultrackingbehaviorsundervariousenvironmentsAseriesofexperiment