基于ADI DSP的高速公路背景建模技术研究与实现
|
摘要
视频检测系统可以实时地收集交通信息数据,以其安装维护便捷、检测面积广等独有优势,逐渐成为了车辆检测领域的主流技术之一。目前,视频车检有几种经典算法,其中背景差法是检测运动目标的有效方法之一。该方法是通过视频当前帧与背景帧进行差分进而得到运动前景,其检测质量主要取决于背景模型的性能。
本文以视频检测系统中的背景建模部分为研究对象,对高速公路背景建模和硬件实现方面进行了研究,在BF 548 DSP评估板上开发了背景建模系统的原型。
在背景建模方面,首先将典型背景建模算法进行对比,并对高速公路视频像素点的灰度特征进行分析总结。然后,以码本方法作为背景建模算法的基础,并对其加以改进。最后,对感兴趣区域进行设置,完成了基于概率统计的背景建模及其更新算法。实验表明,在较少的内核资源使用前提下,该算法拥有较好的背景提取效果,为视频检测系统的后续工作提供了有力的保障。
在硬件实现方面,本文首先分析了当前主流视频检测系统的硬件实现方法。其次,根据背景建模软件算法及其今后拓展的需求,选用BF 548 DSP评估板作为系统的硬件开发平台。然后,将计算机视觉库OpenCV的部分程序移植到BF 548 DSP平台上,并根据Blackfin系列处理器的特点对程序进行优化。最后,在BF 548 DSP评估板上实现了背景建模以及与上位机通信功能。实验结果表明,本文的背景建模系统在满足一定条件下能够准确、实时地对视频进行背景建模,基本达到预期效果。
Video detection system can collect traffic information data on real-time. Because it has some unique advantages such as easy installation and maintenance and the area of testing is wide, video detection system has become to mainstream technique of the field of vehicle detection. At present, there are several classical algorithms for video detection system. Among them, the background subtraction method is an effective way to detect moving targets. This method get moving foreground through the current frame minus the background frame. The quality of test performance depends on background modeling.
The background modeling of the video detection system is used as study object in this paper. Background modeling of highway and its hardware was studied. The prototype of background modeling system was developed on the BF548 DSP evaluation board.
In terms of the background modeling, typical algorithms of background modeling and characteristics of highway were analyzed and summarized. Then, the codebook modeling was used as the basis of the background modeling algorithm and improved. Finally, probability and statistics background modeling and updating algorithms is completed with setting the region of interest. Experiments show that the algorithm could extract background well on using resources of kernel less. That provided a strong guarantee for follow-up the work of video detection system.
In terms of hardware implementation, mainstream hardware implementation of video detection system is analyzed first. Based on the needs of software algorithm of background modeling and future expansion, the BF 548 evaluation board was selected to achieve the algorithm. Then, the parts of the OpenCV were ported to the BF 548 DSP platform. The programs were optimized based on the characteristics of Blackfin family of processors. Finally, achieve to run background modeling algorithm and communication with the host computer on the BF 548 DSP evaluation board. The experimental results show that the background modeling system can accurately get the background in real time under certain conditions. Those achieve the desired effect.
本文以视频检测系统中的背景建模部分为研究对象,对高速公路背景建模和硬件实现方面进行了研究,在BF 548 DSP评估板上开发了背景建模系统的原型。
在背景建模方面,首先将典型背景建模算法进行对比,并对高速公路视频像素点的灰度特征进行分析总结。然后,以码本方法作为背景建模算法的基础,并对其加以改进。最后,对感兴趣区域进行设置,完成了基于概率统计的背景建模及其更新算法。实验表明,在较少的内核资源使用前提下,该算法拥有较好的背景提取效果,为视频检测系统的后续工作提供了有力的保障。
在硬件实现方面,本文首先分析了当前主流视频检测系统的硬件实现方法。其次,根据背景建模软件算法及其今后拓展的需求,选用BF 548 DSP评估板作为系统的硬件开发平台。然后,将计算机视觉库OpenCV的部分程序移植到BF 548 DSP平台上,并根据Blackfin系列处理器的特点对程序进行优化。最后,在BF 548 DSP评估板上实现了背景建模以及与上位机通信功能。实验结果表明,本文的背景建模系统在满足一定条件下能够准确、实时地对视频进行背景建模,基本达到预期效果。
Video detection system can collect traffic information data on real-time. Because it has some unique advantages such as easy installation and maintenance and the area of testing is wide, video detection system has become to mainstream technique of the field of vehicle detection. At present, there are several classical algorithms for video detection system. Among them, the background subtraction method is an effective way to detect moving targets. This method get moving foreground through the current frame minus the background frame. The quality of test performance depends on background modeling.
The background modeling of the video detection system is used as study object in this paper. Background modeling of highway and its hardware was studied. The prototype of background modeling system was developed on the BF548 DSP evaluation board.
In terms of the background modeling, typical algorithms of background modeling and characteristics of highway were analyzed and summarized. Then, the codebook modeling was used as the basis of the background modeling algorithm and improved. Finally, probability and statistics background modeling and updating algorithms is completed with setting the region of interest. Experiments show that the algorithm could extract background well on using resources of kernel less. That provided a strong guarantee for follow-up the work of video detection system.
In terms of hardware implementation, mainstream hardware implementation of video detection system is analyzed first. Based on the needs of software algorithm of background modeling and future expansion, the BF 548 evaluation board was selected to achieve the algorithm. Then, the parts of the OpenCV were ported to the BF 548 DSP platform. The programs were optimized based on the characteristics of Blackfin family of processors. Finally, achieve to run background modeling algorithm and communication with the host computer on the BF 548 DSP evaluation board. The experimental results show that the background modeling system can accurately get the background in real time under certain conditions. Those achieve the desired effect.
引文
[1]张国伍.智能交通系统工程导论[M].第二版.北京：电子工业出版社,2003.9：1-114.
[2]张小军.智能交通系统中的车辆检测和车型识别技术研究[硕士学位论文].重庆：西南大学.2006.6.
[3]魏志强.基于虚拟检测器的视频车辆检测系统研究与实现[硕士学位论文].昆明：昆明理工大学.2010.1.
[4]Barron J, Fleet D, Beauchemin S. Performance of optical flow techniques[J]. International Journal of Computer Vision.1994,12 (1):43-77.
[5]Iketani A, Nagai A, Kuno Y, et al. Real time surveillance system detecting persons in complex scenes[A]. Proceedings of Image Analysis and Processing, IEEE Computer Society. 1999:1112-1115.
[6]Lipton A, Fujiyoshi H, Patil R. Moving target classification and tracking from real-time video[A]. In:Proceedings of IEEE Workshop on Applications of Computer Vision[C], Princeton, USA,1998:8-14.
[7]Anderson C, Bert P, Vander Wal G. Change detection and tracking using pyramids transformation techniques[A]. In:Proceedings of SPIE Conference on Intelligent Robots and Computer Vision[C], Cambridge, MA, USA,1985,579:72～78.
[8]Friedman N, Russell S. Image segmentation in video sequences:A probabilistic approach[C]. In Proc of Conference on Uncertainty in Artificial Intelligence (UAI).1997:175～181.
[9]Stauffer C, WEL Crimson. Adaptive background mixture models for real-time tracking[C]. Proc of the IEEE Conference on Computer Vision and Pattern Recognition,1999,2:246～252.
[10]Elgammal A, Harwood D, Davis L. Non-parametric Model for Background Subtraction[C]. Proceedings of the 6th European Conference on Computer Vision-Part Ⅱ.2000,751～767.
[11]Kim K, Chalidabhongse TH, Harwood D, et al. Background modeling and subtraction by codebook construction[C]. Proceedings of IEEE International Conference on Image Processing. 2004,3061～3064.
[12]Li Y, Xu LQ, Morphett J, et al. An integrated algorithm of incremental and robust pca[C]. Proceedings of IEEE International Conference on Image Processing.2003,245～248.
[13]Heikkila M, Pietikainen M. A texture-based method for modeling the background detecting moving objects[J]. IEEE Transactions on Pattern Analysis and Machine Intelligence.2006,28 (4):657～662.
[14]Heikkila M, Pietikainen M, Heikkila J. A Texture-based Method for Detecting Moving Objects[C]. Proc British Machine Vision Conference.2004:vol.1,187～196.
[15]Ojala T, Pietikainen M, Maenpaa T. Multiresolution gray-scale and rotation invariant texture classification with local binary patterns[J]. IEEE Trans Pattern Analysis and Machine Intelligence.2002,24(7):971～987.
[16]皮小平.基于DSP(DM642)的嵌入式人脸识别器的设计：[硕士学位论文].上海：上海交通大学.2006.
[17]王科俊,李艳波,李国琴.基于DM642的机器人双目视觉系统设计.自动化技术与应用.2006,25(6)：34-36
[18]王光娟,詹永照,刘志强.基于DM642的嵌入式疲劳驾驶监测系统的实现.计算机应用.2007,27(10)：2612-2614
[19]周鄂林.基于DSP的视频监控系统设计及相关算法研究：[硕士学位论文].成都：电子科技大学.2004.
[20]张亚丽.基T-DM642的嵌入式车辆检测系统：[硕士学位论文].浙江：浙江人学.2007
[21]Friedman N, Russell S. Image segmentation in video sequences:A probabilistic approach[C]. In Proc of Conference on Uncertainty in Artificial Intelligence (UAI).1997:175～181.
[22]Mittal A, Huttenlocher D. Scene modeling for wide area surveillance and image synthesis [A]. In:Proceedings of IEEE International Conference On Computer Vision and Pattern Recognition[C], Hilton Head Island, SC, USA,2000:160-167.
[23]Kang J, Cohen I, Medioni G. Continuous tracking within and across camera streams [A]. In: Proceedings of IEEE International Conference on Computer Vision and Pattern Recognition [C], Madison, WI, USA,2003:267～272.
[24]Fuentes L, Velastin S. From tracking to advanced surveillance [A]. In:Proceedings of IEEE International Conference on Image Processing[C], Barcelona Spain,2003:121～124.
[25]Cucchiara R, Piccardi M, Prati A. Detecting moving objects, ghosts, and shadows in video streams[J]. IEEE Transactions on Pattern Analysis and Machine Intelligence,2003,25(10): 1337～1342.
[26]Stauffer C, Grimson W. Learning patterns of activity using real-time tracking[J]. IEEE Transactions on Pattern Analysis and Machine Intelligence,2000,22(8):747～757.
[27]Kaew Tra Kul Pong P, Bowden R. An improved adaptive background mixture model for real-time tracking with shadow detection[A]. In:Proceedings of the 2nd European Workshop on Advanced Video-Based Surveillance Systems[C], Kingston, UK,2001:149～158.
[28]Harville M. A framework for high-level feedback to adaptive, perpixel, mixture-of-Ganssian background models[A]. In:Proceedings of European Conference on Computer Vision[C], Copenhagen, Denmark,2002,3:543～560.
[29]Cucchiara R, Piccardi M, Prati A. Detecting moving objects, ghosts, and shadows in video streams[J]. IEEE Transactions on Pattern Analysis and Machine Intelligence,2003,25(10): 1337～1342.
[30]Elgammal A, Harwood D, Davis L. Non-parametric model for background subtraction[A]. In:Proceedings of International Conference on Computer Vision[c], Kerkyra, Greece,1999: 751～767.
[31]Fuentes L, Velastin S. From tracking to advanced surveillance[A]. In:Proceedings of IEEE International Conference on Image Processing[C], Barcelona Spain,2003:121～124.
[32]Toyama K, Krumm J, Brnmitt B, et al.Wallflower:Principles and practice of background maintenance[A]. In:Proceedings of International Conference on Computer Vision[C], Coffu, Greece,1999:255～261.
[33]Heikkila J, Silven O. A real-time system for monitoring of cyclists and pedestrians[A]. In: Proceedings of IEEE Workshop on Visual Surveillance[C], Fort Collins, Colorado, USA,1999: 246～252.
[34]Haritaoglu I, Harwood D, Davis L. W4:Real-time surveillance of people and their activities[J]. IEEE Transactions on Pattern Analysis and Machine Intelligence,2000,22(8): 809～830.
[35]Gutchess D, A background model initialization algorithm for video surveillance[A]. In: Proceedings of IEEE International Conference on Computer Vision[C], Vancouver, BC, Canada, 2001:744～740.
[36]Prati A, Mikic I, Trivedi M, et al. Detecting moving shadows:algorithms and evaluation[J]. IEEE Transactions on Pattern Analysis and Machine Intelligence,2003,25(7):918～923.
[37]Lipton A, Fujiyoshi H, Patil R. Moving target classification and tracking from real-time video[A]. In:Proceedings of IEEE Workshop on Applications of Computer Vision[C], Princeton, USA,1998:8～14.
[38]Anderson C, Bert P, Vander Wal G. Change detection and tracking using pyramids transformation techniques[A]. In:Proceedings of SPIE Conference on Intelligent Robots and Computer Vision[C], Cambridge, MA, USA,1985,579:72～78.
[39]Collins R, Lipton A J, Kanade T, et al. A system for video surveillance and monitoring: VSAM final report[R]. Technical Report:CMU-RI-TR-00-12, Carnegie Melon University, Pittsburgh, Peen, America,2000.
[40]Ai Hai-zhou, Lv Feng-jun. Changes detection and segmentation in visual surveillance[J]. Computer Engineering and Application,2000,37(5)：75-77.[艾海舟,吕凤军.面向视觉监视的变化检测与分割[J].计算机工程与应用,2000,37(5)：75-77.]
[41]Lin Hong-wen, Tu Dan, Li Guo-hui. Moving objects detection method based on statistical background model[J]. Computer Engineering,2003,29(16):97～99[林洪文,涂丹,李国辉.基于统计背景模型的运动目标检测方法[J].计算机工程,2003,29(16)：97-99.]
[42]Gloyer B, Aghajan H, Siu K Y, et al. Video-based freeway monitoring system using recursive vehicle tracking[A]. In:Proceedings of SPIE Symposium on Electronic Imaging: Image and Video Processing[C], San Jose, CA, USA,1995,2421:173～180.
[43]Lo B, Velastin S. Automatic congestion detection system for underground platforms[A]. In: Proceedings of InternationalSymposium on Intelligent Multimedia, Video, and Speech Processing[C], Hang Kong,2001:158～161.
[44]Zhou Q, Aggarwal J. Tracking and classifying moving objects from videos[A]. In: Proceedings of IEEE Workshop on Performance Evaluation ofTracking and Surveillance[C], Hawaii, USA,2001.
[45]Haritaoglu I, Davis Larry S, Harwood D. W4 who? when? where? what? a real-time system for detecting and tmcking people[A]. In:Proceedings of IEEE International Conference on Automatic Face and Gesture Recognition[C], Nara, Japan,1998:222～227. [46] Halevy G, Weinshall D. Motion of disturbances:detection and tracking of multi-body non-rigid motion[J]. Machine Vision and Applications,1999,11(3):122～137. [47] Keller D, Weber J, Malik J. Robust multiple car tracking with occlusion reasoning[A]. In: Proceedings of European Conference on Computer Vision[C], Stockholm, Sweden,1994: 189～196.
[48]Kdspann K P. Moving object recognition using all adaptive background memory [A]. In: Cappellini V, ed. Time-Varying Image Processing and Moving Object RecognitionfM]. Amsterdam, the Netherlands:Elsevier Science Publishers,1990:289～307.
[49]Kilger M. A shadow handler in a video-based real-time traffic monitoring system[A]. In: Proceedings of IEEE Workshop on Applications of Computer Vision[C], Palm Springs, CA, USA,1992:1060～1066.
[50]Elgammal A. Background and foreground modeling using nonparametric kernel density estimation for visual surveillance[J]. Proceedings of IEEE,2002,90(7):1151～1163.
[51]Friedman N, Russell S. Image segmentation in video sequences:A probabilistic approachf A]. In:Proceedings of the 13"Conference on Uncertainty in Artificial Intelligence[C], Rhode Island, USA,1997:175～181.
[52]Grimson W, Stauffer C, Romano R, Using adaptive tracking to classify and monitor activities in a site[A]. In:Proceedings of IEEE International Conference on Computer Vision and Pattern Recognition [C], Santa Barbara, CA, USA,1998:22～29.
[53]Stauffer C, Grimson W. Adaptive background mixture models for realtime traeking[A]. In: Proceedings of IEEE International Conference on Computer Vision and Pattern Recognition[C], Fort Collins, Colorado, USA,1999,2:246～252.
[54]Gao X, Boult T, Coetzee F, et al. Error analysis of background adaption[A]. In:Proceedings of IEEE International Conference on Computer Vision and Pattern Recognition[C], Hilton Head Isand, SC, USA,2000:503～510.
[55]Power P W, Schoonees J A. Understanding background mixturemodels for foreground segmentation[A]. In:Proceedings of Imageand Vision Computing[C], Auckland, New Zealand, 2002:267～271.
[56]Lee D S, Hull J, Erol B. A Bayesian framework for gaussian mixture background modeling[A]. In:Proceedings of IEEE International Conference on Image Processing[C], Barcelona, Spain,2003:973～976.
[57]Rittscher J, Kate J, Joga S, et al. A probabilistic background model for tracking[A]. In: Proceedings of European Conference on Computer Vision[C], Dublin, Ireland,2000,2: 336～350.
[58]Stenger B, Ramesh V, Paragios N, et al. Topology free hidden markov models:Application to background modeling[A]. In:Proceedings of IEEE International Conference on Computer Vision[C], Vancouver, BC, Canada,2001,1:294～301.
[59]Oliver N, Rosario B, Penfland A. A Bayesian computer vision system for modeling human interactions[J]. IEEE Transactions on Pattern Analysis and Machine Intelligence,2000,22(8): 831～843.
[60]Fukunaga K. Hostetler L D. The estimation of the gradient of a density function, with applications in pattern recognition[J]. IEEE Transactions on Information Theory,1975,21(1): 32～40.
[61]Han B, Comaniciu D, Davis L. Sequential kernel density approximation through mode propagation:applications to background modeling[A]. In:Proceedings of Asian Conference on Computer Vision[c], Jeju Island, Korea,2004.
[62]Yang Chang-jiang, Duraiswami R, De Menthon D.Mean. shift analysis using quasi-newton methods[A]. In:Proceedings of IEEE International Conference on Image Processing[C], Barcelona, Spain,2003,3:447～450.
[63]Kim K, Chalidabhongse T H, Harwood D, et al. Background modeling and subtraction by Codehook construction[A]. In:Proceedings of IEEE International Conference on Image Processing[C], Singapore,2004:3061～3064.
[64]Matsuyahaa T, Ohya T, Habe H. Background subtraction for nonstationary scenes[A]. In: Proceedings of the 4th Asian Conference on Computer Vision[C], Taipel. China,2000:662～667.
[65]Wada T, Matsuyama T. Appearance sphere:Background model for pan-tilt-zoom camera[A]. In:Proceedings of International Conference on Pattern Recognition[c], Vienna, Austria,1996: 718～722.
[66]Cheung S C, Kamath C. Robust techniques for background subtraction in urban traffic video[A]. In:Proceedings of SPIE Electronic Imaging:Visual Communications and Image Processing[C], San Jose, California, USA,2004,1:881～892.
[2]张小军.智能交通系统中的车辆检测和车型识别技术研究[硕士学位论文].重庆：西南大学.2006.6.
[3]魏志强.基于虚拟检测器的视频车辆检测系统研究与实现[硕士学位论文].昆明：昆明理工大学.2010.1.
[4]Barron J, Fleet D, Beauchemin S. Performance of optical flow techniques[J]. International Journal of Computer Vision.1994,12 (1):43-77.
[5]Iketani A, Nagai A, Kuno Y, et al. Real time surveillance system detecting persons in complex scenes[A]. Proceedings of Image Analysis and Processing, IEEE Computer Society. 1999:1112-1115.
[6]Lipton A, Fujiyoshi H, Patil R. Moving target classification and tracking from real-time video[A]. In:Proceedings of IEEE Workshop on Applications of Computer Vision[C], Princeton, USA,1998:8-14.
[7]Anderson C, Bert P, Vander Wal G. Change detection and tracking using pyramids transformation techniques[A]. In:Proceedings of SPIE Conference on Intelligent Robots and Computer Vision[C], Cambridge, MA, USA,1985,579:72～78.
[8]Friedman N, Russell S. Image segmentation in video sequences:A probabilistic approach[C]. In Proc of Conference on Uncertainty in Artificial Intelligence (UAI).1997:175～181.
[9]Stauffer C, WEL Crimson. Adaptive background mixture models for real-time tracking[C]. Proc of the IEEE Conference on Computer Vision and Pattern Recognition,1999,2:246～252.
[10]Elgammal A, Harwood D, Davis L. Non-parametric Model for Background Subtraction[C]. Proceedings of the 6th European Conference on Computer Vision-Part Ⅱ.2000,751～767.
[11]Kim K, Chalidabhongse TH, Harwood D, et al. Background modeling and subtraction by codebook construction[C]. Proceedings of IEEE International Conference on Image Processing. 2004,3061～3064.
[12]Li Y, Xu LQ, Morphett J, et al. An integrated algorithm of incremental and robust pca[C]. Proceedings of IEEE International Conference on Image Processing.2003,245～248.
[13]Heikkila M, Pietikainen M. A texture-based method for modeling the background detecting moving objects[J]. IEEE Transactions on Pattern Analysis and Machine Intelligence.2006,28 (4):657～662.
[14]Heikkila M, Pietikainen M, Heikkila J. A Texture-based Method for Detecting Moving Objects[C]. Proc British Machine Vision Conference.2004:vol.1,187～196.
[15]Ojala T, Pietikainen M, Maenpaa T. Multiresolution gray-scale and rotation invariant texture classification with local binary patterns[J]. IEEE Trans Pattern Analysis and Machine Intelligence.2002,24(7):971～987.
[16]皮小平.基于DSP(DM642)的嵌入式人脸识别器的设计：[硕士学位论文].上海：上海交通大学.2006.
[17]王科俊,李艳波,李国琴.基于DM642的机器人双目视觉系统设计.自动化技术与应用.2006,25(6)：34-36
[18]王光娟,詹永照,刘志强.基于DM642的嵌入式疲劳驾驶监测系统的实现.计算机应用.2007,27(10)：2612-2614
[19]周鄂林.基于DSP的视频监控系统设计及相关算法研究：[硕士学位论文].成都：电子科技大学.2004.
[20]张亚丽.基T-DM642的嵌入式车辆检测系统：[硕士学位论文].浙江：浙江人学.2007
[21]Friedman N, Russell S. Image segmentation in video sequences:A probabilistic approach[C]. In Proc of Conference on Uncertainty in Artificial Intelligence (UAI).1997:175～181.
[22]Mittal A, Huttenlocher D. Scene modeling for wide area surveillance and image synthesis [A]. In:Proceedings of IEEE International Conference On Computer Vision and Pattern Recognition[C], Hilton Head Island, SC, USA,2000:160-167.
[23]Kang J, Cohen I, Medioni G. Continuous tracking within and across camera streams [A]. In: Proceedings of IEEE International Conference on Computer Vision and Pattern Recognition [C], Madison, WI, USA,2003:267～272.
[24]Fuentes L, Velastin S. From tracking to advanced surveillance [A]. In:Proceedings of IEEE International Conference on Image Processing[C], Barcelona Spain,2003:121～124.
[25]Cucchiara R, Piccardi M, Prati A. Detecting moving objects, ghosts, and shadows in video streams[J]. IEEE Transactions on Pattern Analysis and Machine Intelligence,2003,25(10): 1337～1342.
[26]Stauffer C, Grimson W. Learning patterns of activity using real-time tracking[J]. IEEE Transactions on Pattern Analysis and Machine Intelligence,2000,22(8):747～757.
[27]Kaew Tra Kul Pong P, Bowden R. An improved adaptive background mixture model for real-time tracking with shadow detection[A]. In:Proceedings of the 2nd European Workshop on Advanced Video-Based Surveillance Systems[C], Kingston, UK,2001:149～158.
[28]Harville M. A framework for high-level feedback to adaptive, perpixel, mixture-of-Ganssian background models[A]. In:Proceedings of European Conference on Computer Vision[C], Copenhagen, Denmark,2002,3:543～560.
[29]Cucchiara R, Piccardi M, Prati A. Detecting moving objects, ghosts, and shadows in video streams[J]. IEEE Transactions on Pattern Analysis and Machine Intelligence,2003,25(10): 1337～1342.
[30]Elgammal A, Harwood D, Davis L. Non-parametric model for background subtraction[A]. In:Proceedings of International Conference on Computer Vision[c], Kerkyra, Greece,1999: 751～767.
[31]Fuentes L, Velastin S. From tracking to advanced surveillance[A]. In:Proceedings of IEEE International Conference on Image Processing[C], Barcelona Spain,2003:121～124.
[32]Toyama K, Krumm J, Brnmitt B, et al.Wallflower:Principles and practice of background maintenance[A]. In:Proceedings of International Conference on Computer Vision[C], Coffu, Greece,1999:255～261.
[33]Heikkila J, Silven O. A real-time system for monitoring of cyclists and pedestrians[A]. In: Proceedings of IEEE Workshop on Visual Surveillance[C], Fort Collins, Colorado, USA,1999: 246～252.
[34]Haritaoglu I, Harwood D, Davis L. W4:Real-time surveillance of people and their activities[J]. IEEE Transactions on Pattern Analysis and Machine Intelligence,2000,22(8): 809～830.
[35]Gutchess D, A background model initialization algorithm for video surveillance[A]. In: Proceedings of IEEE International Conference on Computer Vision[C], Vancouver, BC, Canada, 2001:744～740.
[36]Prati A, Mikic I, Trivedi M, et al. Detecting moving shadows:algorithms and evaluation[J]. IEEE Transactions on Pattern Analysis and Machine Intelligence,2003,25(7):918～923.
[37]Lipton A, Fujiyoshi H, Patil R. Moving target classification and tracking from real-time video[A]. In:Proceedings of IEEE Workshop on Applications of Computer Vision[C], Princeton, USA,1998:8～14.
[38]Anderson C, Bert P, Vander Wal G. Change detection and tracking using pyramids transformation techniques[A]. In:Proceedings of SPIE Conference on Intelligent Robots and Computer Vision[C], Cambridge, MA, USA,1985,579:72～78.
[39]Collins R, Lipton A J, Kanade T, et al. A system for video surveillance and monitoring: VSAM final report[R]. Technical Report:CMU-RI-TR-00-12, Carnegie Melon University, Pittsburgh, Peen, America,2000.
[40]Ai Hai-zhou, Lv Feng-jun. Changes detection and segmentation in visual surveillance[J]. Computer Engineering and Application,2000,37(5)：75-77.[艾海舟,吕凤军.面向视觉监视的变化检测与分割[J].计算机工程与应用,2000,37(5)：75-77.]
[41]Lin Hong-wen, Tu Dan, Li Guo-hui. Moving objects detection method based on statistical background model[J]. Computer Engineering,2003,29(16):97～99[林洪文,涂丹,李国辉.基于统计背景模型的运动目标检测方法[J].计算机工程,2003,29(16)：97-99.]
[42]Gloyer B, Aghajan H, Siu K Y, et al. Video-based freeway monitoring system using recursive vehicle tracking[A]. In:Proceedings of SPIE Symposium on Electronic Imaging: Image and Video Processing[C], San Jose, CA, USA,1995,2421:173～180.
[43]Lo B, Velastin S. Automatic congestion detection system for underground platforms[A]. In: Proceedings of InternationalSymposium on Intelligent Multimedia, Video, and Speech Processing[C], Hang Kong,2001:158～161.
[44]Zhou Q, Aggarwal J. Tracking and classifying moving objects from videos[A]. In: Proceedings of IEEE Workshop on Performance Evaluation ofTracking and Surveillance[C], Hawaii, USA,2001.
[45]Haritaoglu I, Davis Larry S, Harwood D. W4 who? when? where? what? a real-time system for detecting and tmcking people[A]. In:Proceedings of IEEE International Conference on Automatic Face and Gesture Recognition[C], Nara, Japan,1998:222～227. [46] Halevy G, Weinshall D. Motion of disturbances:detection and tracking of multi-body non-rigid motion[J]. Machine Vision and Applications,1999,11(3):122～137. [47] Keller D, Weber J, Malik J. Robust multiple car tracking with occlusion reasoning[A]. In: Proceedings of European Conference on Computer Vision[C], Stockholm, Sweden,1994: 189～196.
[48]Kdspann K P. Moving object recognition using all adaptive background memory [A]. In: Cappellini V, ed. Time-Varying Image Processing and Moving Object RecognitionfM]. Amsterdam, the Netherlands:Elsevier Science Publishers,1990:289～307.
[49]Kilger M. A shadow handler in a video-based real-time traffic monitoring system[A]. In: Proceedings of IEEE Workshop on Applications of Computer Vision[C], Palm Springs, CA, USA,1992:1060～1066.
[50]Elgammal A. Background and foreground modeling using nonparametric kernel density estimation for visual surveillance[J]. Proceedings of IEEE,2002,90(7):1151～1163.
[51]Friedman N, Russell S. Image segmentation in video sequences:A probabilistic approachf A]. In:Proceedings of the 13"Conference on Uncertainty in Artificial Intelligence[C], Rhode Island, USA,1997:175～181.
[52]Grimson W, Stauffer C, Romano R, Using adaptive tracking to classify and monitor activities in a site[A]. In:Proceedings of IEEE International Conference on Computer Vision and Pattern Recognition [C], Santa Barbara, CA, USA,1998:22～29.
[53]Stauffer C, Grimson W. Adaptive background mixture models for realtime traeking[A]. In: Proceedings of IEEE International Conference on Computer Vision and Pattern Recognition[C], Fort Collins, Colorado, USA,1999,2:246～252.
[54]Gao X, Boult T, Coetzee F, et al. Error analysis of background adaption[A]. In:Proceedings of IEEE International Conference on Computer Vision and Pattern Recognition[C], Hilton Head Isand, SC, USA,2000:503～510.
[55]Power P W, Schoonees J A. Understanding background mixturemodels for foreground segmentation[A]. In:Proceedings of Imageand Vision Computing[C], Auckland, New Zealand, 2002:267～271.
[56]Lee D S, Hull J, Erol B. A Bayesian framework for gaussian mixture background modeling[A]. In:Proceedings of IEEE International Conference on Image Processing[C], Barcelona, Spain,2003:973～976.
[57]Rittscher J, Kate J, Joga S, et al. A probabilistic background model for tracking[A]. In: Proceedings of European Conference on Computer Vision[C], Dublin, Ireland,2000,2: 336～350.
[58]Stenger B, Ramesh V, Paragios N, et al. Topology free hidden markov models:Application to background modeling[A]. In:Proceedings of IEEE International Conference on Computer Vision[C], Vancouver, BC, Canada,2001,1:294～301.
[59]Oliver N, Rosario B, Penfland A. A Bayesian computer vision system for modeling human interactions[J]. IEEE Transactions on Pattern Analysis and Machine Intelligence,2000,22(8): 831～843.
[60]Fukunaga K. Hostetler L D. The estimation of the gradient of a density function, with applications in pattern recognition[J]. IEEE Transactions on Information Theory,1975,21(1): 32～40.
[61]Han B, Comaniciu D, Davis L. Sequential kernel density approximation through mode propagation:applications to background modeling[A]. In:Proceedings of Asian Conference on Computer Vision[c], Jeju Island, Korea,2004.
[62]Yang Chang-jiang, Duraiswami R, De Menthon D.Mean. shift analysis using quasi-newton methods[A]. In:Proceedings of IEEE International Conference on Image Processing[C], Barcelona, Spain,2003,3:447～450.
[63]Kim K, Chalidabhongse T H, Harwood D, et al. Background modeling and subtraction by Codehook construction[A]. In:Proceedings of IEEE International Conference on Image Processing[C], Singapore,2004:3061～3064.
[64]Matsuyahaa T, Ohya T, Habe H. Background subtraction for nonstationary scenes[A]. In: Proceedings of the 4th Asian Conference on Computer Vision[C], Taipel. China,2000:662～667.
[65]Wada T, Matsuyama T. Appearance sphere:Background model for pan-tilt-zoom camera[A]. In:Proceedings of International Conference on Pattern Recognition[c], Vienna, Austria,1996: 718～722.
[66]Cheung S C, Kamath C. Robust techniques for background subtraction in urban traffic video[A]. In:Proceedings of SPIE Electronic Imaging:Visual Communications and Image Processing[C], San Jose, California, USA,2004,1:881～892.