目标跟踪与背景减除算法研究
|
摘要
计算机视觉在最近几年中得到了很大的发展。然而计算机视觉领域仍有许多问题有待解决,其中背景减除与目标跟踪是目前研究者关注的焦点。背景减除和目标跟踪算法是视频监控、人机交互以及车辆导航等许多应用的关键技术之一。近年来,人们已经在背景减除和目标跟踪方面取得重要进展,提出了许多重要的背景减除以及目标跟踪算法。然而,背景减除和目标跟踪是一个非常困难的问题,更加鲁棒可靠的背景减除以及目标跟踪算法仍有待进一步发掘。本文对常见的背景减除及目标跟踪算法进行了研究与改进,提出了几种新的背景减除与目标跟踪算法。
在背景减除中,混合高斯模型是一种较常见的模型。由于标准的EM算法是离线算法,因此虽然EM算法可以较为准确的估计GMM模型的参数,却无法用于在线背景减除应用中。为解决这个问题,本文提出了一种新的基于充分统计量的在线EM算法。实验结果表明这种算法可以高效准确的估计出GMM模型的参数。
K-Means算法属于一种聚类算法,具有实现简单,运行效率高的特点。广义K-Means算法是对K-Means算法的扩展,能够估计出GMM模型中的所有参数。通过分析EM算法和K-Means算法之间的联系,本文提出了一种基于充分统计量的在线K-Means算法。实验显示这种算法的性能同基于Robbin-Monro近似的在线K-Means算法性能接近。
本文基于以上算法对视频进行了背景减除实验,结果表明所提出的在线EM算法和在线K-Means算法都能够可靠的检测到运动目标。为了对算法的性能进行详细的分析和比较,本文采用一维及二维仿真数据对上述算法的性能进行了实验对比。
高斯粒子滤波是贝叶斯滤波的一种半参数化的实现,已经被人们成功的用于目标跟踪中。本文提出了一种新的采样算法。通过将高斯粒子滤波的采样及预测步骤合并为一步,高斯粒子滤波的实现得到了简化。本文证明当系统的预测模型为线性时,新的采样步骤同原采样及预测步骤等价。当系统的预测模型为非线性时,可以先通过一阶泰勒展开对非线性模型线性化,然后再应用本文的采样方法。
基于模板匹配算法的目标跟踪中,在遮挡或目标变形情况下,会出现跟踪不稳定的现象。为解决这个问题,本文提出了一种新的似然值度量函数。对比实验表明,改进后的模板匹配算法可以稳定的跟踪到运动目标
最后本文在背景减除的基础上做了目标跟踪实验。实验结果表明背景减除提高了目标跟踪的稳定性。
Although significant progress has been made in the field of computer vision, a lot of important problems in this field still need to be resolved. Among them, background subtraction and object tacking has received many researcher's attention in recent years. Background subtraction and object tracking is one of the fundamental tasks in many ap-plications, such as visual surveillance, human computer interaction and vehicle navigation. Significant progress has been made in background subtraction and object tracking during the last few years, and several important algorithms have been developed for solving this problem. However, due to the fact that background subtraction and object tracking is a challenging problem, more efficient and robust background subtraction and object track-ing algorithm is still need to be explored. In this dissertation, some popular algorithms in background subtraction and object tracking are studied and several new algorithms are proposed.
Gaussian Mixture Models is one of the most popular models for background sub-traction. Although EM algorithm can estimate GMM parameters accurately, it cannot be applied to background subtraction directly. The main reason is that EM is not an incre-mental algorithm. To solve this problem, a new sufficient statistic vector based online EM algorithm is proposed. Experimental results demonstrate that our approach can effectively and efficiently estimate GMM parameters.
K-Means algorithm is a cluster algorithm which is simple to implement and efficient to run. Elliptical K-Means is an extension of K-Means and can also be used to calculate GMM parameters. By exploring the close relationship between EM and K-Means, a sufficient statistic vector based online K-Means algorithm is presented. Experimental results show that the performance of the new online K-Means algorithm is comparable with that of the Robbins-Monro approximated online K-Means algorithm.
The usefulness of the proposed two online algorithms in background subtraction is validated on real video data and the performance of them is analyzed using simulated data.
Gaussian Particle filter is a semi-parameter realization of Bayes filter and has been successfully applied to object tracking application. To simplify the realization of GPF, we proposed a new particle sampling method which combines the sampling step with state prediction step by taking advantage of the Gaussian assumption and by exploring the linear structure of the system dynamic model. We proved that when the system dynamic model is linear, the new sampling method is equal to the combination of the original sampling step and the state prediction step. When the system dynamic model is nonlinear, we use Taylor method to approximate the nonlinear model with a linear model. The proposed sampling method can then be applied to the approximated linear model.
To solve the stability problem of template based object tracking algorithm, we pro-posed a new similarity measure function and applied it to an adaptive template matching based object tracking algorithm. Comparative experiments demonstrate that the improved algorithm can track object stably even serious occlusion present.
Finally we demonstrate that the performance of object tracking can be improved greatly by combining background subtraction as a preprocessing step.
在背景减除中,混合高斯模型是一种较常见的模型。由于标准的EM算法是离线算法,因此虽然EM算法可以较为准确的估计GMM模型的参数,却无法用于在线背景减除应用中。为解决这个问题,本文提出了一种新的基于充分统计量的在线EM算法。实验结果表明这种算法可以高效准确的估计出GMM模型的参数。
K-Means算法属于一种聚类算法,具有实现简单,运行效率高的特点。广义K-Means算法是对K-Means算法的扩展,能够估计出GMM模型中的所有参数。通过分析EM算法和K-Means算法之间的联系,本文提出了一种基于充分统计量的在线K-Means算法。实验显示这种算法的性能同基于Robbin-Monro近似的在线K-Means算法性能接近。
本文基于以上算法对视频进行了背景减除实验,结果表明所提出的在线EM算法和在线K-Means算法都能够可靠的检测到运动目标。为了对算法的性能进行详细的分析和比较,本文采用一维及二维仿真数据对上述算法的性能进行了实验对比。
高斯粒子滤波是贝叶斯滤波的一种半参数化的实现,已经被人们成功的用于目标跟踪中。本文提出了一种新的采样算法。通过将高斯粒子滤波的采样及预测步骤合并为一步,高斯粒子滤波的实现得到了简化。本文证明当系统的预测模型为线性时,新的采样步骤同原采样及预测步骤等价。当系统的预测模型为非线性时,可以先通过一阶泰勒展开对非线性模型线性化,然后再应用本文的采样方法。
基于模板匹配算法的目标跟踪中,在遮挡或目标变形情况下,会出现跟踪不稳定的现象。为解决这个问题,本文提出了一种新的似然值度量函数。对比实验表明,改进后的模板匹配算法可以稳定的跟踪到运动目标
最后本文在背景减除的基础上做了目标跟踪实验。实验结果表明背景减除提高了目标跟踪的稳定性。
Although significant progress has been made in the field of computer vision, a lot of important problems in this field still need to be resolved. Among them, background subtraction and object tacking has received many researcher's attention in recent years. Background subtraction and object tracking is one of the fundamental tasks in many ap-plications, such as visual surveillance, human computer interaction and vehicle navigation. Significant progress has been made in background subtraction and object tracking during the last few years, and several important algorithms have been developed for solving this problem. However, due to the fact that background subtraction and object tracking is a challenging problem, more efficient and robust background subtraction and object track-ing algorithm is still need to be explored. In this dissertation, some popular algorithms in background subtraction and object tracking are studied and several new algorithms are proposed.
Gaussian Mixture Models is one of the most popular models for background sub-traction. Although EM algorithm can estimate GMM parameters accurately, it cannot be applied to background subtraction directly. The main reason is that EM is not an incre-mental algorithm. To solve this problem, a new sufficient statistic vector based online EM algorithm is proposed. Experimental results demonstrate that our approach can effectively and efficiently estimate GMM parameters.
K-Means algorithm is a cluster algorithm which is simple to implement and efficient to run. Elliptical K-Means is an extension of K-Means and can also be used to calculate GMM parameters. By exploring the close relationship between EM and K-Means, a sufficient statistic vector based online K-Means algorithm is presented. Experimental results show that the performance of the new online K-Means algorithm is comparable with that of the Robbins-Monro approximated online K-Means algorithm.
The usefulness of the proposed two online algorithms in background subtraction is validated on real video data and the performance of them is analyzed using simulated data.
Gaussian Particle filter is a semi-parameter realization of Bayes filter and has been successfully applied to object tracking application. To simplify the realization of GPF, we proposed a new particle sampling method which combines the sampling step with state prediction step by taking advantage of the Gaussian assumption and by exploring the linear structure of the system dynamic model. We proved that when the system dynamic model is linear, the new sampling method is equal to the combination of the original sampling step and the state prediction step. When the system dynamic model is nonlinear, we use Taylor method to approximate the nonlinear model with a linear model. The proposed sampling method can then be applied to the approximated linear model.
To solve the stability problem of template based object tracking algorithm, we pro-posed a new similarity measure function and applied it to an adaptive template matching based object tracking algorithm. Comparative experiments demonstrate that the improved algorithm can track object stably even serious occlusion present.
Finally we demonstrate that the performance of object tracking can be improved greatly by combining background subtraction as a preprocessing step.
引文
[1]朱淼良.计算机视觉[M]. 1st ed浙江,杭州：浙江大学出版社,1997.
[2]Gonzalez R C, Woods R E数字图像处理[M].2nd ed北京：电子工业出版社,2002.
[3]Jain R C, Nagel H H. On the analysis of accumulative difference pictures from image sequences of real world scenes[J]. IEEE Transactions on Pattern Analysis and Machine Intelligence,1979, 1(2):206-214.
[4]Pentland A P, Darrell T J, Azarbayejani A, et al. Pfinder:Real-time tracking of the human body[J].1996,51-56.
[5]Gao D, Mahadevan V, Vasconcelos N. On the plausibility of the discriminant center-surround hypothesis for visual saliency.[J]. Journal of vision,2008,8(7).
[6]Stauffer C, Grimson W E L. Learning patterns of activity using real-time tracking[J]. IEEE Trans. Pattern Anal. Mach. Intell,2000,22(8):747-757.
[7]Elgammal A, Harwood D, Davis L. Non-parametric model for background subtrac-tion 2000.
[8]Li L Y, Leung M K H. Integrating intensity and texture differences for robust change detection[J]. IEEE Trans. Image Processing,2002,11(2):105-112.
[9]Toyama K, Krumm J, Brumitt B, et al. Wallflower:Principles and practice of back-ground maintenance [C]. ICCV.1999:255-261.
[10]Rittscher J, Kato J, Joga S, et al. A probabilistic background model for tracking[C]. Vernon D. ECCV (2). vol 1843. Springer 2000:336-350.
[11]Stenger B, Ramesh V, Paragios N, et al. Topology free hidden markov models: Application to background modeling[C]. ICCV.2001:Ⅰ:294-301.
[12]Oliver N M, Rosario B, Pentland A P. A bayesian computer vision system for mod-eling human interactions[J]. IEEE Transactions on Pattern Analysis and Machine Intelligence,2000,22(8):831-843.
[13]Kanade T, Collins R T, Lipton A J, et al. Advances in cooperative multi-sensor video surveillance[C]. Image Understanding Workshop.1998:3-24.
[14]Rowe S M, Blake A. Statistical mosaics for tracking[J]. Image and Vision Comput-ing,1996,14(8):549-564.
[15]Irani M, Anandan P. Video indexing based on mosaic representations [J]. Proceed-ings of IEEE,1998,86(5):905-921.
[16]Monnet A, Mittal A, Paragios N, et al. Background modeling and subtraction of dynamic scenes[C]. ICCV.2003:1305-1312.
[17]Zhong J; Sclaroff S. Segmenting foreground objects from a dynamic textured back-ground via a robust kalman filter[C]. ICCV.2003:44-50.
[18]Grill-SlBIec K, Valappil J. The human visual cortex [J]. Annual Review of Neuro-science,12004,27:647-677.
[19]Kanwisher N, Wojciulik E. Visual attention:Insights from brain imaging[J]. Nature Reviews Neuroscience,2000,1:91-100.
[20]Kanwisher N, Wojciulik E. A bottom-up model of spatial attention predicts human error patterns in rapid scene recognition[J]. Journal of Vision,2007,7:1-13.
[21]Dashan G, Vasconcelos N. Bottom-up saliency is a discriminant process[C]. ICCV. 2007:1-6.
[22]Gao D, Han S, Vasconcelos N. Discriminant saliency, the detection of suspicious co-incidences, and applications to visual recognition[J]. IEEE Transactions on Pattern Analysis and Machine Intelligence,2009,31(6):989-1005.
[23]Peters R, Itti L. Beyond bottom-up:Incorporating task-dependent influences into a computational model of spatial attention[C]. CVPR.2007:1-8.
[24]Mahadevan V, Vasconcelos N. Background subtraction in highly dynamic scenes[C]. CVPR.2008:1-6.
[25]Mahadevan V, Vasconcelos N. Saliency-based discriminant tracking[C]. CVPR. 2009:1007-1013.
[26]Alper Y, Omar J, Mubarak S. Object tracking:A survey[J]. ACM Comput. Surv., 2006,38(4):1-13.
[27]Sethi I, Jain R. Finding trajectories of feature points in a monocular image se-quence[J]. IEEE Trans. Patt. Analy. Mach. Intell,1987,9:56-73.
[28]Salari V, Sethi I K. Feature point correspondence in the presence of occlusion[J]. IEEE Trans. Patt. Analy. Mach. Intell,1990,12:87-91.
[29]Rangarajan K, Shah M. Establishing motion correspondence[C]. Conference Vision Graphies Image Process.1991:56-73.
[30]Shafique K, Shah M. A non-iterative greedy algorithm for multi-frame point cor-respondence[C]. In IEEE International Conference on Computer Vision (ICCV). 2003:110-115.
[31]Broida T, Chellappa R. Estimation of object motion parameters from noisy im-ages[J]. IEEE Trans. Patt. Analy. Mach. Intell,1986,12:90-99.
[32]Rosales R, Sclaroff S.3d trajectory recovery for tracking multiple objects and tra-jectory guided recognition of actions[C]. In IEEE Conference on Computer Vision and Pattern Recognition (CVPR).1999:117-123.
[33]Tanizaki H. Non-gaussian state-space modeling of nonstationary time series[J]. J. Amer. Statist. Assoc.,1987,82:1032-1063.
[34]Chang Y L, Aggarwal J K.3d structure reconstruction from an ego motion sequence using statistical estimation and detection theory[C]. In Workshop on Visual Motion. 1991:268-273.
[35]Rasmussen C, Hanger G. Probabilistic data association methods for tracking com-plex visual objects[J]. IEEE Trans. Patt. Analy. Mach. Intell,2001,23:560-576.
[36]Cham T,Rehg J M. A multiple hypothesis approach to figure tracking[C]. In IEEE Conference on Computer Vision and Pattern Recognition (CVPR).1999:239-245.
[37]Schweitzer H, Bell J W, Wu F. Very fast template matching[C]. In European Con-ference on Computer Vision (ECCV).2002:358-372.
[38]Comaniciu D, Ramesh V, Meer P. Kernel-based object tracking[J]. IEEE Transac-tions on Pattern Analysis and Machine Intelligence,2003,25:564-577.
[39]Horn B, Schunk B. Determining optical flow[J]. Artific. Intell.,1981,17:185-203.
[40]Shi J, Tomasi C. Good features to track[C]. In IEEE Conference on Computer Vision and Pattern Recognition (CVPR).1994:593-600.
[41]Tao H, Sawhney H, Kumar R. Object tracking with bayesian estimation of dynamic layer representations[J]. IEEE Trans. Patt. Analy. Mach. Intell,2002,24:75-89.
[42]Isard M, Maccormick J. Bramble:A bayesian multiple-blob tracker[C]. In IEEE International Conference on Computer Vision (ICCV).2001:34-41.
[43]Black M, Jepson A. Eigentracking:Robust matching and tracking of articulated objects using a view-based representation[J]. Int. J. Comput. Vision,1998,26:63-84.
[44]Avidan S. Support vector tracking[C]. In IEEE Conference on Computer Vision and Pattern Recognition (CVPR).2001:184-191.
[45]Huttenlocher D, Noh J, Rucklldge W. Tracking nonrigid objects in complex scenes[C]. In IEEE International Conference on Computer Vision (ICCV).1993: 93-101.
[46]Li B, Chellappa R, Zheng Q, et al. Model-based temporal object verification using video[J]. IEEE Trans. Image Process,2001,10:897-908.
[47]Kang J, Cohen I, Medioni G. Object reacquisition using geometric invariant appear-ance model[C]. In International Conference on Pattern Recongnition (ICPR).2004: 759-762.
[48]Sato K, Aggarwal J. Temporal spatio-velocity transform and its application to track-ing and interaction[J]. Comput. Vision Image Understand,2004,96:100-128.
[49]Terzopoulos D, Szellski R. Tracking with kalman snakes 1992.
[50]Isard M, Blake A. Condensation-conditional density propagation for visual track-ing[J]. Int. J. Comput. Vision,1998,29:5-28.
[51]Maccormick J, Blake A. Probabilistic exclusion and partitioned sampling for multi-ple object tracking[J]. Int. J. Comput. Vision,2000,39:57-71.
[52]Chen Y, Rui Y, Huang T. Jpdaf based hmm for real-time contour tracking[C]. In IEEE Conference on Computer Vision and Pattern Recognition (CVPR).2001:543-550.
[53]Bertalmio M, Sapiro G, Randall G. Morphing active contours[J]. IEEE Trans. Patt. Analy. Mach. Intell,2000,22:733-737.
[54]Ronfard R. Region based strategies for active contour models[J]. Int. J. Comput. Vision,1994,13:229-251.
[55]Yilmaz A, Shafique K, Shah M. Target tracking in airborne forward looking im-agery[J]. J.Image Vision Comput,2003,21:623-635.
[56]王江涛.基于视频的目标检测、跟踪及其行为识别研究[D].博士学位论文.南京：南京理工大学,2008.
[57]文志强.基于均值偏移算法的运动目标跟踪技术的研究[D].博士学位论文.长沙：中南大学,2008.
[58]张玲.视频目标跟踪方法研究[D].博士学位论文.合肥：中国科技大学,2009.
[59]王书朋.视频目标跟踪算法研究[D].博士学位论文.西安：西安电子科技大学,2009.
[60]明安龙.智能视觉监控中目标跟踪与识别算法研究[D].博士学位论文.北京：北京邮电大学,2008.
[61]李铮.智能视频监控中的遮挡目标跟踪技术研究[D].博士学位论文.武汉：华中科技大学,2008.
[62]Schon T. Estimation of nonlinear dynamic systems theory and applications 2006.
[63]Thrun S, Burgard W, Fox D. Probabilistic robotics 2005.
[64]Jazwinski A. Stochastic processes and filtering theory[M].1st ed. New York, USA: Academic Press,1970.
[65]Anderson B, Moore J. Optimal Filtering[M]. 1st ed. Englewood Cliffs, NJ, USA: Prentice Hall,1979.
[66]Kailath T, Sayed A H, Hassibi B. Linear Estimation[M]. 1st ed. Upper Saddle River, NJ, USA:Prentice Hall,2000.
[67]Bar-Shalom Y, Fortmann T. Tracking and Data Association[M]. 1st ed. Orlando, FL, USA:Academic Press,1988.
[68]Gustafsson F. Adaptive Filtering and Change Detection[M].1st ed. New York, USA:John Wiley Sons,2000.
[69]Metropolis N, Ulam S. The monte carlo method[J]. Journal of the American Statis-tical Association,1949,44:335-341.
[70]Hammersley J, Morton K. Poor man's monte carlo[J]. Journal of the Royal Statisti-cal Socity B,1954,16:23-38.
[71]Doucet A, Godsill S, Andrieu C. On sequential monte carlo sampling methods for bayesian filtering 1998.
[72]Rubin D. Using the SIR algorithm to simulate posterior distributions[M].1st ed. Oxford, UK:Oxford University Press,1988.
[73]Gordon N, Salmond D, Smith A. Novel approcach to nonlinear/non-gaussian bayesian state estimation[J]. In IEE Proceedings on Radar and Signal Processing, 1993,140:107-113.
[74]Liu J, Chen E. Sequential monte carlo methods for dynamics systems[J]. Journal of American Statistical Association,1998,93:1032-1044.
[75]Robert C, Casella G. Monte Carlo Statistical Methods[M].1st ed. New York, USA: Springer,1999.
[76]Doucet A, Godsill S, Andrieu C. On sequential monte carlo sampling methods for bayesian filtring[J]. Statistics and Computing,2000,10:197-208.
[77]Arulampalam M, Maskell S, Gordon N, et al. A tutorial on particle filters for online nonlinear/non-gaussian bayesian tracking[J]. IEEE Transactions on Signal Process-ing,2002,50:174-188.
[78]Schon T, Gustafsson F. Particle filters for system identification of statespace models linear in either parameters or states [C]. In Proceedings of the 13 th IFAC Symposium on System Identification.2003:1287-1292.
[79]Dempster A, Laird N, Rubin D. Maximum likelihood from incomplete data via the em algorithm[J]. Journal of the Royal Statistical Socity B,1977,39:1-38.
[80]Mclachlan G, Krishnan T. The em algorithm and its extensions 1997.
[81]Csiszar I, Tusnady G. Information geometry and alternating minimization proce-dures[J]. Statistics and Decisions,1984,1:205-237.
[82]Hathaway R. Another interpretation of the em algorithm for mixture distributions[J]. Statistics and Probability Letters,1986,4:53-56.
[83]Neal R, Hinton G. A new view of the em algorithm that justifies incremental and other variants 1999.
[84]Nowlan S J. Soft competitive Adaptation:Neural Network Learning Algorithms :Neural Network Learning Algorithms based on Fitting Statistical Mixtures[D]. PhD thesis. School of Computer Science, Carnegie Mellon University, Pittsburgh.,1991.
[85]文志强,蔡自兴Mean-shift算法的收敛性分析[J].软件学报,2007,18：205-212.
[86]Kotecha J H, Djuric P A. Gaussian particle filtering [J]. IEEE Transactions on Signal Processing,2003,51(1-10):2592-2601.
[87]Ghirmai T. Gaussian particle filtering for tracking maneuvering targets [J]. Proceed-ings. IEEE SoutheastCon 2007 (IEEE Cat. No.07CH37882),2007,22-25.
[88]WU X d, SONG Z h. Gaussian particle filter based pose and motion estimation[J]. LZhejiang Univ. Sci. A,2007,8(1-10):1604-1613. [89] Sorenson H. Least-squares estimation:from gauss to kalman[J]. IEEE Spectrum, 1970,7:63-68.[90]张金国,吕庭豪.基于matlab的改进模板匹配方法的机器人目标识别系统[J].机器人,2003,25：623-625.[91]姜映映,田丰,王旭刚,et al基于模板匹配和svm的草图符号自适应识别方法[J].计算机学报,2009,32：252-260.
[92]张彦梅,宋扬.模板匹配方法在高速目标跟踪中的应用[J].北京理工大学学报,2006,26：1026-1029.
[93]宋仁庭,杨卫平,杨明月.模板匹配算法对运动目标自动锁定跟踪的研究[J].红外与激光工程,2007,36：197-200.
[94]黄飞,李德华,姚迅.基于相关匹配及自适应模板更新的目标跟踪新方法[J].计算机工程,2007,33：147-179.
[95]Kyriacou. Theocharis. Guido B, Stanislao L. Vision-based urban navigation proce-dures for verbally instructed robots[J]. Robotics and Autonomous Systems,2005, 51:69-80.
[96]Ross D A, Lim J, Lin R S, et al. Incremental learning for robust visual tracking[J]. International Journal of Computer Vision,2008,77:125-141.
[97]陈皓,马彩文,陈岳承,et al基于灰度统计的快速模板匹配算法[J].光子学报,2009,38：1586-1590.
[98]季艳,张英慧.基于小波变换的模板匹配方法研究[J].中国科技大学学报,2008,38：293-296.
[99]Xu X Y. Li B X. Adaptive rao-blackwellized particle filter and its evaluation for tracking in surveillance[J]. Ieee Transactions on Image Processing,2007, 16(3):838-849.
[100]Duan Z H, Cai Z X, Yu J X. An adaptive particle filter for soft fault compensa-tion of mobile robots [J]. Science in China Series F-Information Sciences,2008, 51(12):2033-2046.
[101]Fox D. Adapting the sample size in particle filters through kld-sampling[J]. Inter-national Journal of Robotics Research,2003,22(12):985-1003.
[102]Stauffer C, Grimson W. Adaptive background mixture models for real-time track-ing[C]. ICCV.1999:246-252.
[103]Zivkovic Z, Heijden F v d. Efficient adaptive density estimation per image pixel for the task of background subtraction[J]. Pattern Recognition Letters,2006,27(7):773-780.
[104]Koller D, Weber J, Huang T, et al. Towards robust automatic traffic scene analysis in real-time[C]. ICPR.1994:126-131.
[105]Wayne P, Schoonees J. Understanding background mixture models for foreground segmentation[C]. Proc. of IVCNZ,2002:267-271.
[106]Howe N R, Drechamps A. Better foreground segmentation through graph cuts[J]. Computing Research Repository,2004, cs.CV:1-8.
[107]Jeff A B. A gentle tutorial of the EM algorithm and its application to parameter estimation for gaussian mixture and hidden markov models[R]. Berkeley, California, USA:International Computer Science Institute,1998.
[108]Zivkovic Z, van der Heijden F. Efficient adaptive density estimation per image pixel for the task of background subtraction[J]. Pattern Recognition Letters,2006, 27(7):773-780.
[109]Bishop C M. Pattern Recognition and Machine Learning[M].1st ed. Cambridge: Springer,2008.
[110]Hedvig S, Michael J, Leonid S. Implicit probabilistic models of human motion for synthesis and tracking[C]. European Conference on Computer Vision.2002: 784-800.
[2]Gonzalez R C, Woods R E数字图像处理[M].2nd ed北京：电子工业出版社,2002.
[3]Jain R C, Nagel H H. On the analysis of accumulative difference pictures from image sequences of real world scenes[J]. IEEE Transactions on Pattern Analysis and Machine Intelligence,1979, 1(2):206-214.
[4]Pentland A P, Darrell T J, Azarbayejani A, et al. Pfinder:Real-time tracking of the human body[J].1996,51-56.
[5]Gao D, Mahadevan V, Vasconcelos N. On the plausibility of the discriminant center-surround hypothesis for visual saliency.[J]. Journal of vision,2008,8(7).
[6]Stauffer C, Grimson W E L. Learning patterns of activity using real-time tracking[J]. IEEE Trans. Pattern Anal. Mach. Intell,2000,22(8):747-757.
[7]Elgammal A, Harwood D, Davis L. Non-parametric model for background subtrac-tion 2000.
[8]Li L Y, Leung M K H. Integrating intensity and texture differences for robust change detection[J]. IEEE Trans. Image Processing,2002,11(2):105-112.
[9]Toyama K, Krumm J, Brumitt B, et al. Wallflower:Principles and practice of back-ground maintenance [C]. ICCV.1999:255-261.
[10]Rittscher J, Kato J, Joga S, et al. A probabilistic background model for tracking[C]. Vernon D. ECCV (2). vol 1843. Springer 2000:336-350.
[11]Stenger B, Ramesh V, Paragios N, et al. Topology free hidden markov models: Application to background modeling[C]. ICCV.2001:Ⅰ:294-301.
[12]Oliver N M, Rosario B, Pentland A P. A bayesian computer vision system for mod-eling human interactions[J]. IEEE Transactions on Pattern Analysis and Machine Intelligence,2000,22(8):831-843.
[13]Kanade T, Collins R T, Lipton A J, et al. Advances in cooperative multi-sensor video surveillance[C]. Image Understanding Workshop.1998:3-24.
[14]Rowe S M, Blake A. Statistical mosaics for tracking[J]. Image and Vision Comput-ing,1996,14(8):549-564.
[15]Irani M, Anandan P. Video indexing based on mosaic representations [J]. Proceed-ings of IEEE,1998,86(5):905-921.
[16]Monnet A, Mittal A, Paragios N, et al. Background modeling and subtraction of dynamic scenes[C]. ICCV.2003:1305-1312.
[17]Zhong J; Sclaroff S. Segmenting foreground objects from a dynamic textured back-ground via a robust kalman filter[C]. ICCV.2003:44-50.
[18]Grill-SlBIec K, Valappil J. The human visual cortex [J]. Annual Review of Neuro-science,12004,27:647-677.
[19]Kanwisher N, Wojciulik E. Visual attention:Insights from brain imaging[J]. Nature Reviews Neuroscience,2000,1:91-100.
[20]Kanwisher N, Wojciulik E. A bottom-up model of spatial attention predicts human error patterns in rapid scene recognition[J]. Journal of Vision,2007,7:1-13.
[21]Dashan G, Vasconcelos N. Bottom-up saliency is a discriminant process[C]. ICCV. 2007:1-6.
[22]Gao D, Han S, Vasconcelos N. Discriminant saliency, the detection of suspicious co-incidences, and applications to visual recognition[J]. IEEE Transactions on Pattern Analysis and Machine Intelligence,2009,31(6):989-1005.
[23]Peters R, Itti L. Beyond bottom-up:Incorporating task-dependent influences into a computational model of spatial attention[C]. CVPR.2007:1-8.
[24]Mahadevan V, Vasconcelos N. Background subtraction in highly dynamic scenes[C]. CVPR.2008:1-6.
[25]Mahadevan V, Vasconcelos N. Saliency-based discriminant tracking[C]. CVPR. 2009:1007-1013.
[26]Alper Y, Omar J, Mubarak S. Object tracking:A survey[J]. ACM Comput. Surv., 2006,38(4):1-13.
[27]Sethi I, Jain R. Finding trajectories of feature points in a monocular image se-quence[J]. IEEE Trans. Patt. Analy. Mach. Intell,1987,9:56-73.
[28]Salari V, Sethi I K. Feature point correspondence in the presence of occlusion[J]. IEEE Trans. Patt. Analy. Mach. Intell,1990,12:87-91.
[29]Rangarajan K, Shah M. Establishing motion correspondence[C]. Conference Vision Graphies Image Process.1991:56-73.
[30]Shafique K, Shah M. A non-iterative greedy algorithm for multi-frame point cor-respondence[C]. In IEEE International Conference on Computer Vision (ICCV). 2003:110-115.
[31]Broida T, Chellappa R. Estimation of object motion parameters from noisy im-ages[J]. IEEE Trans. Patt. Analy. Mach. Intell,1986,12:90-99.
[32]Rosales R, Sclaroff S.3d trajectory recovery for tracking multiple objects and tra-jectory guided recognition of actions[C]. In IEEE Conference on Computer Vision and Pattern Recognition (CVPR).1999:117-123.
[33]Tanizaki H. Non-gaussian state-space modeling of nonstationary time series[J]. J. Amer. Statist. Assoc.,1987,82:1032-1063.
[34]Chang Y L, Aggarwal J K.3d structure reconstruction from an ego motion sequence using statistical estimation and detection theory[C]. In Workshop on Visual Motion. 1991:268-273.
[35]Rasmussen C, Hanger G. Probabilistic data association methods for tracking com-plex visual objects[J]. IEEE Trans. Patt. Analy. Mach. Intell,2001,23:560-576.
[36]Cham T,Rehg J M. A multiple hypothesis approach to figure tracking[C]. In IEEE Conference on Computer Vision and Pattern Recognition (CVPR).1999:239-245.
[37]Schweitzer H, Bell J W, Wu F. Very fast template matching[C]. In European Con-ference on Computer Vision (ECCV).2002:358-372.
[38]Comaniciu D, Ramesh V, Meer P. Kernel-based object tracking[J]. IEEE Transac-tions on Pattern Analysis and Machine Intelligence,2003,25:564-577.
[39]Horn B, Schunk B. Determining optical flow[J]. Artific. Intell.,1981,17:185-203.
[40]Shi J, Tomasi C. Good features to track[C]. In IEEE Conference on Computer Vision and Pattern Recognition (CVPR).1994:593-600.
[41]Tao H, Sawhney H, Kumar R. Object tracking with bayesian estimation of dynamic layer representations[J]. IEEE Trans. Patt. Analy. Mach. Intell,2002,24:75-89.
[42]Isard M, Maccormick J. Bramble:A bayesian multiple-blob tracker[C]. In IEEE International Conference on Computer Vision (ICCV).2001:34-41.
[43]Black M, Jepson A. Eigentracking:Robust matching and tracking of articulated objects using a view-based representation[J]. Int. J. Comput. Vision,1998,26:63-84.
[44]Avidan S. Support vector tracking[C]. In IEEE Conference on Computer Vision and Pattern Recognition (CVPR).2001:184-191.
[45]Huttenlocher D, Noh J, Rucklldge W. Tracking nonrigid objects in complex scenes[C]. In IEEE International Conference on Computer Vision (ICCV).1993: 93-101.
[46]Li B, Chellappa R, Zheng Q, et al. Model-based temporal object verification using video[J]. IEEE Trans. Image Process,2001,10:897-908.
[47]Kang J, Cohen I, Medioni G. Object reacquisition using geometric invariant appear-ance model[C]. In International Conference on Pattern Recongnition (ICPR).2004: 759-762.
[48]Sato K, Aggarwal J. Temporal spatio-velocity transform and its application to track-ing and interaction[J]. Comput. Vision Image Understand,2004,96:100-128.
[49]Terzopoulos D, Szellski R. Tracking with kalman snakes 1992.
[50]Isard M, Blake A. Condensation-conditional density propagation for visual track-ing[J]. Int. J. Comput. Vision,1998,29:5-28.
[51]Maccormick J, Blake A. Probabilistic exclusion and partitioned sampling for multi-ple object tracking[J]. Int. J. Comput. Vision,2000,39:57-71.
[52]Chen Y, Rui Y, Huang T. Jpdaf based hmm for real-time contour tracking[C]. In IEEE Conference on Computer Vision and Pattern Recognition (CVPR).2001:543-550.
[53]Bertalmio M, Sapiro G, Randall G. Morphing active contours[J]. IEEE Trans. Patt. Analy. Mach. Intell,2000,22:733-737.
[54]Ronfard R. Region based strategies for active contour models[J]. Int. J. Comput. Vision,1994,13:229-251.
[55]Yilmaz A, Shafique K, Shah M. Target tracking in airborne forward looking im-agery[J]. J.Image Vision Comput,2003,21:623-635.
[56]王江涛.基于视频的目标检测、跟踪及其行为识别研究[D].博士学位论文.南京：南京理工大学,2008.
[57]文志强.基于均值偏移算法的运动目标跟踪技术的研究[D].博士学位论文.长沙：中南大学,2008.
[58]张玲.视频目标跟踪方法研究[D].博士学位论文.合肥：中国科技大学,2009.
[59]王书朋.视频目标跟踪算法研究[D].博士学位论文.西安：西安电子科技大学,2009.
[60]明安龙.智能视觉监控中目标跟踪与识别算法研究[D].博士学位论文.北京：北京邮电大学,2008.
[61]李铮.智能视频监控中的遮挡目标跟踪技术研究[D].博士学位论文.武汉：华中科技大学,2008.
[62]Schon T. Estimation of nonlinear dynamic systems theory and applications 2006.
[63]Thrun S, Burgard W, Fox D. Probabilistic robotics 2005.
[64]Jazwinski A. Stochastic processes and filtering theory[M].1st ed. New York, USA: Academic Press,1970.
[65]Anderson B, Moore J. Optimal Filtering[M]. 1st ed. Englewood Cliffs, NJ, USA: Prentice Hall,1979.
[66]Kailath T, Sayed A H, Hassibi B. Linear Estimation[M]. 1st ed. Upper Saddle River, NJ, USA:Prentice Hall,2000.
[67]Bar-Shalom Y, Fortmann T. Tracking and Data Association[M]. 1st ed. Orlando, FL, USA:Academic Press,1988.
[68]Gustafsson F. Adaptive Filtering and Change Detection[M].1st ed. New York, USA:John Wiley Sons,2000.
[69]Metropolis N, Ulam S. The monte carlo method[J]. Journal of the American Statis-tical Association,1949,44:335-341.
[70]Hammersley J, Morton K. Poor man's monte carlo[J]. Journal of the Royal Statisti-cal Socity B,1954,16:23-38.
[71]Doucet A, Godsill S, Andrieu C. On sequential monte carlo sampling methods for bayesian filtering 1998.
[72]Rubin D. Using the SIR algorithm to simulate posterior distributions[M].1st ed. Oxford, UK:Oxford University Press,1988.
[73]Gordon N, Salmond D, Smith A. Novel approcach to nonlinear/non-gaussian bayesian state estimation[J]. In IEE Proceedings on Radar and Signal Processing, 1993,140:107-113.
[74]Liu J, Chen E. Sequential monte carlo methods for dynamics systems[J]. Journal of American Statistical Association,1998,93:1032-1044.
[75]Robert C, Casella G. Monte Carlo Statistical Methods[M].1st ed. New York, USA: Springer,1999.
[76]Doucet A, Godsill S, Andrieu C. On sequential monte carlo sampling methods for bayesian filtring[J]. Statistics and Computing,2000,10:197-208.
[77]Arulampalam M, Maskell S, Gordon N, et al. A tutorial on particle filters for online nonlinear/non-gaussian bayesian tracking[J]. IEEE Transactions on Signal Process-ing,2002,50:174-188.
[78]Schon T, Gustafsson F. Particle filters for system identification of statespace models linear in either parameters or states [C]. In Proceedings of the 13 th IFAC Symposium on System Identification.2003:1287-1292.
[79]Dempster A, Laird N, Rubin D. Maximum likelihood from incomplete data via the em algorithm[J]. Journal of the Royal Statistical Socity B,1977,39:1-38.
[80]Mclachlan G, Krishnan T. The em algorithm and its extensions 1997.
[81]Csiszar I, Tusnady G. Information geometry and alternating minimization proce-dures[J]. Statistics and Decisions,1984,1:205-237.
[82]Hathaway R. Another interpretation of the em algorithm for mixture distributions[J]. Statistics and Probability Letters,1986,4:53-56.
[83]Neal R, Hinton G. A new view of the em algorithm that justifies incremental and other variants 1999.
[84]Nowlan S J. Soft competitive Adaptation:Neural Network Learning Algorithms :Neural Network Learning Algorithms based on Fitting Statistical Mixtures[D]. PhD thesis. School of Computer Science, Carnegie Mellon University, Pittsburgh.,1991.
[85]文志强,蔡自兴Mean-shift算法的收敛性分析[J].软件学报,2007,18：205-212.
[86]Kotecha J H, Djuric P A. Gaussian particle filtering [J]. IEEE Transactions on Signal Processing,2003,51(1-10):2592-2601.
[87]Ghirmai T. Gaussian particle filtering for tracking maneuvering targets [J]. Proceed-ings. IEEE SoutheastCon 2007 (IEEE Cat. No.07CH37882),2007,22-25.
[88]WU X d, SONG Z h. Gaussian particle filter based pose and motion estimation[J]. LZhejiang Univ. Sci. A,2007,8(1-10):1604-1613. [89] Sorenson H. Least-squares estimation:from gauss to kalman[J]. IEEE Spectrum, 1970,7:63-68.[90]张金国,吕庭豪.基于matlab的改进模板匹配方法的机器人目标识别系统[J].机器人,2003,25：623-625.[91]姜映映,田丰,王旭刚,et al基于模板匹配和svm的草图符号自适应识别方法[J].计算机学报,2009,32：252-260.
[92]张彦梅,宋扬.模板匹配方法在高速目标跟踪中的应用[J].北京理工大学学报,2006,26：1026-1029.
[93]宋仁庭,杨卫平,杨明月.模板匹配算法对运动目标自动锁定跟踪的研究[J].红外与激光工程,2007,36：197-200.
[94]黄飞,李德华,姚迅.基于相关匹配及自适应模板更新的目标跟踪新方法[J].计算机工程,2007,33：147-179.
[95]Kyriacou. Theocharis. Guido B, Stanislao L. Vision-based urban navigation proce-dures for verbally instructed robots[J]. Robotics and Autonomous Systems,2005, 51:69-80.
[96]Ross D A, Lim J, Lin R S, et al. Incremental learning for robust visual tracking[J]. International Journal of Computer Vision,2008,77:125-141.
[97]陈皓,马彩文,陈岳承,et al基于灰度统计的快速模板匹配算法[J].光子学报,2009,38：1586-1590.
[98]季艳,张英慧.基于小波变换的模板匹配方法研究[J].中国科技大学学报,2008,38：293-296.
[99]Xu X Y. Li B X. Adaptive rao-blackwellized particle filter and its evaluation for tracking in surveillance[J]. Ieee Transactions on Image Processing,2007, 16(3):838-849.
[100]Duan Z H, Cai Z X, Yu J X. An adaptive particle filter for soft fault compensa-tion of mobile robots [J]. Science in China Series F-Information Sciences,2008, 51(12):2033-2046.
[101]Fox D. Adapting the sample size in particle filters through kld-sampling[J]. Inter-national Journal of Robotics Research,2003,22(12):985-1003.
[102]Stauffer C, Grimson W. Adaptive background mixture models for real-time track-ing[C]. ICCV.1999:246-252.
[103]Zivkovic Z, Heijden F v d. Efficient adaptive density estimation per image pixel for the task of background subtraction[J]. Pattern Recognition Letters,2006,27(7):773-780.
[104]Koller D, Weber J, Huang T, et al. Towards robust automatic traffic scene analysis in real-time[C]. ICPR.1994:126-131.
[105]Wayne P, Schoonees J. Understanding background mixture models for foreground segmentation[C]. Proc. of IVCNZ,2002:267-271.
[106]Howe N R, Drechamps A. Better foreground segmentation through graph cuts[J]. Computing Research Repository,2004, cs.CV:1-8.
[107]Jeff A B. A gentle tutorial of the EM algorithm and its application to parameter estimation for gaussian mixture and hidden markov models[R]. Berkeley, California, USA:International Computer Science Institute,1998.
[108]Zivkovic Z, van der Heijden F. Efficient adaptive density estimation per image pixel for the task of background subtraction[J]. Pattern Recognition Letters,2006, 27(7):773-780.
[109]Bishop C M. Pattern Recognition and Machine Learning[M].1st ed. Cambridge: Springer,2008.
[110]Hedvig S, Michael J, Leonid S. Implicit probabilistic models of human motion for synthesis and tracking[C]. European Conference on Computer Vision.2002: 784-800.