資源描述:
《基于多貝努利濾波多目標(biāo)跟蹤算法研究》由會(huì)員上傳分享,免費(fèi)在線閱讀,更多相關(guān)內(nèi)容在學(xué)術(shù)論文-天天文庫(kù)。
1、本科畢業(yè)論文設(shè)計(jì)基于多貝努利濾波的多目標(biāo)跟蹤算法研究2018年04月06日基于多貝努利濾波的多目標(biāo)跟蹤算法研究摘要:多目標(biāo)跟蹤的目的是從受噪聲和雜波干擾的量測(cè)集中聯(lián)合估計(jì)多目標(biāo)狀態(tài)和目標(biāo)航跡。近年來(lái),各種傳感器快速發(fā)展,目標(biāo)跟蹤技術(shù)已被廣泛地應(yīng)用于工程實(shí)際問(wèn)題中。然而,目標(biāo)不確定性和量測(cè)不確定性給多目標(biāo)跟蹤算法的實(shí)際應(yīng)用帶來(lái)了困難。針對(duì)上述問(wèn)題,本文重點(diǎn)研究了基于MeMBer(Multi-TargetMulti-Bernoulli)濾波的多目標(biāo)跟蹤方法。在文中介紹了原始的MeMBer濾波算法,然而在原始MeMBer濾波算法中目標(biāo)數(shù)目上
2、存在顯著偏差。針對(duì)這一問(wèn)題提出了勢(shì)平衡多目標(biāo)多貝努利(CBMeMBer)濾波,有效的解決了這個(gè)問(wèn)題。此為,提出了高斯混合和序貫蒙特卡洛兩種方式實(shí)現(xiàn)勢(shì)平衡多貝努利濾波算法。實(shí)驗(yàn)結(jié)果表明,在低雜波和檢測(cè)概率高的情況下,序貫蒙特卡洛實(shí)現(xiàn)CBMeMBer濾波器的性能優(yōu)于CPHD和PHD濾波器,而高斯混合實(shí)現(xiàn)CPHD濾波器的性能仍然優(yōu)越,CBMeMBer濾波器只能達(dá)到PHD濾波器類(lèi)似的性能。關(guān)鍵詞:隨機(jī)集,多目標(biāo)跟蹤,多貝努力,蒙特卡諾Multi-TargetTrackingAlgorithmBasedonMulti-BernoulliFilt
3、erAuthor:WangYuedongTutor:HaoZhengqingAbstract:Thepurposeofmulti-targettrackingistojointlyestimatemulti-targetStatesandtargettracksfromtheconcentrationofnoiseandclutter.Inrecentyears,varioussensorsaredevelopingrapidly,andtargettrackingtechnologyhasbeenwidelyappliedineng
4、ineeringpractice.However,theuncertaintyoftargetandmeasurementuncertaintybringdifficultiestothepracticalapplicationofmulti-targettrackingalgorithm.Inviewoftheaboveproblems,thispaperfocusesonmulti-targettrackingmethodbasedonMeMBer(Multi-TargetMulti-Bernoulli)filtering.The
5、originalMeMBerfilteringalgorithmisintroducedinthispaper.However,therearesignificantdeviationsinthenumberoftargetsintheoriginalMeMBerfilteringalgorithm.Inordertosolvethisproblem,apotentialequilibriummultitarget(CBMeMBer)filterisproposed,whicheffectivelysolvesthisproblem.
6、ThisistoproposeGaussequilibriumandsequentialMonteCarlotwowaystorealizeCBMeMBerfilteringalgorithm.Theexperimentalresultsshowthatinthecaseoflowclutterandhighdetectionprobability,theperformanceofthesequentialMonteCarloCBMeMBerfilterissuperiortothatofCPHDandPHDfilters,while
7、theperformanceoftheGaussfilterisstillsuperiortotheCPHDfilter,andtheCBMeMBerfiltercanonlyachievethesimilarabilityofthePHDfilter.Keywords:randomfiniteset,Multi-TargetTracking,Multi-Bernoulli,MonteCarlo目錄1緒論………………………11.1研究背景及意義………………………11.2論文主要內(nèi)容及章節(jié)安排………………………22介紹隨機(jī)有限集及多貝努
8、利濾波算法………………………32.1引言………………………32.2隨機(jī)有限集………………………32.2.1隨機(jī)有限集定義………………………32.2.2多目標(biāo)貝葉斯遞推………………………32.3多目標(biāo)多貝努力濾波算法………