資源描述：
《基于小波變換的粒子濾波目標(biāo)跟蹤算法.pdf》由會(huì)員上傳分享，免費(fèi)在線閱讀，更多相關(guān)內(nèi)容在行業(yè)資料-天天文庫(kù)。
1、第40卷第2期東南大學(xué)學(xué)報(bào)(自然科學(xué)版)Vo.l40No.22010年3月JOURNALOFSOUTHEASTUNIVERSITY(NaturalScienceEdition)Mar.2010doi:10.3969/.jissn.1001-0505.2010.02.020基于小波變換的粒子濾波目標(biāo)跟蹤算法1,213章飛周杏鵬陳小惠1(東南大學(xué)復(fù)雜工程系統(tǒng)測(cè)量與控制教育部重點(diǎn)實(shí)驗(yàn)室,南京210096)2(江蘇科技大學(xué)電子信息學(xué)院,鎮(zhèn)江212003)3(南京郵電大學(xué)自動(dòng)化學(xué)院,南京210046)摘要:針對(duì)純方位被動(dòng)目標(biāo)跟蹤中粒子濾波算法固有的計(jì)算復(fù)雜性問(wèn)題,提出了一種基于小波變
2、換的粒子濾波算法(WMPF).對(duì)粒子權(quán)重進(jìn)行小波多分辨率分解,通過(guò)設(shè)定閾值對(duì)高通部分的粒子權(quán)重進(jìn)行濾波,再根據(jù)重構(gòu)后的粒子權(quán)重去掉重復(fù)粒子,生成新的粒子集來(lái)近似后驗(yàn)概率密度函數(shù),從而在保證濾波精度的同時(shí)大量減少粒子數(shù),提高粒子濾波的計(jì)算效率.將WMPF算法與標(biāo)準(zhǔn)粒子濾波算法應(yīng)用于具有非線性非高斯特點(diǎn)的純方位目標(biāo)跟蹤問(wèn)題,仿真結(jié)果表明,WMPF算法的跟蹤精度與標(biāo)準(zhǔn)粒子濾波算法相當(dāng),計(jì)算效率卻遠(yuǎn)高于標(biāo)準(zhǔn)粒子濾波算法,增強(qiáng)了跟蹤的實(shí)時(shí)性,并且該算法有望進(jìn)一步擴(kuò)展粒子濾波的應(yīng)用范圍.關(guān)鍵詞:粒子濾波;小波變換;多分辨率;非線性濾波;純方位中圖分類號(hào):TP274文獻(xiàn)標(biāo)志碼:A文章編號(hào)
3、:1001-0505(2010)0220320206Particlefiltertargettrackingalgorithmbasedonwavelettransform1,213ZhangFeiZhouXingpengChenXiaohui(1KeyLaboratoryofMeasurementandControlofCSEofMinistryofEducation,SoutheastUniversity,Nanjing210096,China)(2SchoolofElectronicsandInformation,JiangsuUniversityofSciencea
4、ndTechnology,Zhenjiang212003,China)(3SchoolofAutomation,NanjingUniversityofPostsandTelecommunications,Nanjing210046,China)Abstract:Focusingonthenaturalcomputationalcomplexityproblemofparticlefilterinbearings2onlypassivetargettrackingproblem,anewparticlefilterbasedonwavelettransformispropos
5、ed.Waveletmultiresolutiondecompositioniscarriedoutonparticleweights.Bysettingthresholdtofilterthehighpassedparticleweights,thereconstructedparticleweightsareusedtoremovetherepeatedparticleandgenerateanewparticlesettoapproximateposteriorprobabilitydensityfunction.Therefore,thenumberofpartic
6、lesisreducedandthecomputationalefficiencyisimprovedwhilemaintainingfilteringaccuracy.Theproposedparticlefilterisutilizedtosolvethebearings2onlytargetmotionanalysisproblemwiththenonlinearandnon2Gaussiancharacteris2tics,andtomakeacomparisonintrackingefficiencywithstandardparticlefilter.Simul
7、ationresultsdemonstratethatcomparedwithstandardparticlefilteralgorithm,theproposedalgorithmhascomparabletrackingaccuracyandoutclassedcomputationalefficiency,andhasenhancedreal2timetrackingperformance.Moreover,theproposedalgorithmispromisinginexpandingtheapplic
