發(fā)布時(shí)間：2018-11-27 18:29

【摘要】：車(chē)輛軌跡識(shí)別與在線(xiàn)異常檢測(cè)是智能交通系統(tǒng)中的一個(gè)重要研究方向,它對(duì)現(xiàn)實(shí)生活中發(fā)生的交通事故及時(shí)有效地救援和后期處理具有很大的輔助作用;同時(shí)它能減少因?yàn)榻煌ㄊ鹿室�起的交通延誤和二次事故;它還可以為城市交通監(jiān)控與安全管理提供重要的信息憑證。通常情況下,車(chē)輛軌跡識(shí)別與在線(xiàn)異常檢測(cè)是基于軌跡建模的方法進(jìn)行的,而軌跡建模方法分為統(tǒng)計(jì)模型和運(yùn)動(dòng)模型兩種。無(wú)監(jiān)督學(xué)習(xí)是近幾年提出的一種新的軌跡建模方法,該方法能在訓(xùn)練數(shù)據(jù)中排除異常軌跡,并且在訓(xùn)練數(shù)據(jù)比較多的情況下有效地進(jìn)行軌跡識(shí)別及異常檢測(cè),但是對(duì)于正常收集的含有異常軌跡的初始軌跡集及含有少量軌跡的初始軌跡集,該方法的識(shí)別率及異常檢測(cè)精度比較低。針對(duì)這些問(wèn)題,本文將增量式(incremental)EM算法應(yīng)用到無(wú)監(jiān)督軌跡建模上,提出一種基于批處理(batch-mode)模型初始化的增量式軌跡建模方法,并將其應(yīng)用到車(chē)輛軌跡識(shí)別與在線(xiàn)異常檢測(cè)上。首先采用改進(jìn)的Hausdorff距離測(cè)量初始軌跡集中軌跡之間的相似度,再用譜聚類(lèi)算法對(duì)初始軌跡集進(jìn)行聚類(lèi),提取軌跡的分布模式。對(duì)于初始軌跡集中聚類(lèi)完成的每一類(lèi)樣本軌跡,利用多觀(guān)察序列訓(xùn)練方法建立每類(lèi)軌跡的隱馬爾科夫模型,依此來(lái)得到初始軌跡模型庫(kù)。對(duì)于從視頻圖像中提取的新軌跡,使用最大后驗(yàn)估計(jì)尋找新軌跡最可能的正常軌跡類(lèi),再用自動(dòng)閾值法進(jìn)行在線(xiàn)異常檢測(cè),然后識(shí)別新軌跡所屬的軌跡模型,通過(guò)incremental EM算法更新新軌跡所屬類(lèi)別的隱馬爾科夫模型參數(shù)。為了增加本文方法的適應(yīng)性,最后需要對(duì)模型結(jié)構(gòu)進(jìn)行更新。應(yīng)用本文方法對(duì)戶(hù)外實(shí)際場(chǎng)景拍攝視頻進(jìn)行測(cè)試,實(shí)驗(yàn)結(jié)果表明,與經(jīng)典的無(wú)監(jiān)督批處理軌跡建模算法相比,本文方法即增量式軌跡建�？梢缘玫礁�加準(zhǔn)確的軌跡模型庫(kù)、更快的運(yùn)算速度;該算法在初始軌跡集包含異常軌跡的情況下能大幅度提高軌跡識(shí)別率;在異常檢測(cè)方面得到了更高的檢測(cè)率和更低的虛警率,實(shí)現(xiàn)了在線(xiàn)異常檢測(cè),具有對(duì)初始軌跡集不敏感的特點(diǎn)。
[Abstract]:Vehicle trajectory identification and on-line anomaly detection are important research directions in intelligent transportation system. They can help traffic accidents in real life to rescue and deal with them in a timely and effective manner. At the same time, it can reduce traffic delays and secondary accidents caused by traffic accidents, and it can also provide important information evidence for urban traffic monitoring and safety management. Generally, vehicle trajectory identification and online anomaly detection are based on trajectory modeling, and trajectory modeling methods are divided into statistical model and motion model. Unsupervised learning is a new trajectory modeling method proposed in recent years. This method can eliminate abnormal trajectory in training data, and can effectively identify and detect the trajectory when there are more training data. However, the recognition rate and the detection accuracy of the method are low for the normal collection of initial trajectory sets with abnormal trajectories and the initial trajectory sets with a small number of tracks. To solve these problems, this paper applies incremental (incremental) EM algorithm to unsupervised trajectory modeling, and proposes an incremental trajectory modeling method based on batch processing (batch-mode) model initialization. It is applied to vehicle track recognition and online anomaly detection. Firstly, the improved Hausdorff distance is used to measure the similarity between the trajectories of the initial locus, and then the spectral clustering algorithm is used to cluster the initial locus to extract the distribution pattern of the locus. The hidden Markov model of each kind of trajectory is established by using the training method of multiple observation sequences for each kind of sample locus which is clustered by the initial locus, and the initial locus model library is obtained according to this model. For the new trajectory extracted from the video image, the maximum a posteriori estimation is used to find the most probable normal trajectory class, and then the automatic threshold method is used for on-line anomaly detection, and then the trajectory model to which the new trajectory belongs is identified. The hidden Markov model parameters of the new trajectory are updated by incremental EM algorithm. In order to increase the adaptability of this method, we need to update the model structure at last. The experiment results show that, compared with the classical unsupervised batch trajectory modeling algorithm, the incremental trajectory modeling method in this paper can get more accurate trajectory model base. Faster computing speed; The algorithm can greatly improve the trajectory recognition rate when the initial trajectory set contains abnormal trajectory. Higher detection rate and lower false alarm rate are obtained in anomaly detection, and on-line anomaly detection is realized, which is insensitive to the initial trajectory set.
【學(xué)位授予單位】：哈爾濱工程大學(xué)
【學(xué)位級(jí)別】：碩士
【學(xué)位授予年份】：2014
【分類(lèi)號(hào)】：U495;TP391.41

【參考文獻(xiàn)】

相關(guān)期刊論文 前8條

1 梁浩哲;黃魁華;李國(guó)輝;張軍;;基于運(yùn)動(dòng)相似性的監(jiān)控軌跡聚合分析[J];國(guó)防科技大學(xué)學(xué)報(bào);2011年05期

2 曲琳;周凡;陳耀武;;基于Hausdorff距離的視覺(jué)監(jiān)控軌跡分類(lèi)算法[J];吉林大學(xué)學(xué)報(bào)(工學(xué)版);2009年06期

3 郝久月;李超;高磊;熊璋;;智能監(jiān)控場(chǎng)景中運(yùn)動(dòng)目標(biāo)軌跡聚類(lèi)算法[J];北京航空航天大學(xué)學(xué)報(bào);2009年09期

4 袁和金;張艷寧;周濤;佘紅偉;李秀秀;;基于歸一化編輯距離和譜聚類(lèi)的軌跡模式學(xué)習(xí)方法[J];計(jì)算機(jī)輔助設(shè)計(jì)與圖形學(xué)學(xué)報(bào);2008年06期

5 潘奇明;程詠梅;楊濤;潘泉;趙春暉;;真實(shí)場(chǎng)景運(yùn)動(dòng)目標(biāo)軌跡有效性判斷與自動(dòng)聚類(lèi)算法研究[J];計(jì)算機(jī)應(yīng)用研究;2007年04期

6 李和平;胡占義;吳毅紅;吳福朝;;基于半監(jiān)督學(xué)習(xí)的行為建模與異常檢測(cè)[J];軟件學(xué)報(bào);2007年03期

7 裴繼紅,李翠蕓,龔忻;一種新的隱馬爾可夫模型及其在手繪圖形識(shí)別中的應(yīng)用[J];計(jì)算機(jī)學(xué)報(bào);2005年10期

8 胡衛(wèi)明,謝丹,譚鐵牛,沈俊;軌跡分布模式學(xué)習(xí)的層次自組織神經(jīng)網(wǎng)絡(luò)方法[J];計(jì)算機(jī)學(xué)報(bào);2003年04期

相關(guān)博士學(xué)位論文 前3條

1 李嵩松;基于隱馬爾可夫模型和計(jì)算智能的股票價(jià)格時(shí)間序列預(yù)測(cè)[D];哈爾濱工業(yè)大學(xué);2011年

2 胡宏宇;基于視頻處理的交通事件識(shí)別方法研究[D];吉林大學(xué);2010年

3 聞帆;基于視覺(jué)的交通路口車(chē)輛智能檢測(cè)技術(shù)研究[D];哈爾濱工業(yè)大學(xué);2010年

相關(guān)碩士學(xué)位論文 前4條

1 羅懷金;基于近鄰路徑的自適應(yīng)尺度譜聚類(lèi)算法研究[D];哈爾濱工程大學(xué);2012年

2 白巖;基于運(yùn)動(dòng)目標(biāo)軌跡分析的智能交通監(jiān)控系統(tǒng)[D];中山大學(xué);2011年

3 孫望;語(yǔ)音識(shí)別技術(shù)的研究及其在發(fā)音錯(cuò)誤識(shí)別系統(tǒng)中的應(yīng)用[D];南京航空航天大學(xué);2008年

4 潘奇明;運(yùn)動(dòng)目標(biāo)軌跡分類(lèi)與識(shí)別研究[D];西北工業(yè)大學(xué);2006年

，

本文編號(hào)：2361676