基于视频运动检测的多车道交通流信息采集技术研究
|
摘要
随着现代经济的高速发展,交通问题已成为众所周知的热点和难点问题。将计算机科学与自动化等高新技术运用于交通监控管理与车辆控制,以保障交通顺畅及行车安全,从而改善环境质量,促进经济和谐发展的智能交通系统(Intelligent traffic system, ITS)也随之应运而生。在智能交通管理系统中,实时获取交通流信息是智能交通管理系统中重要的一环。利用视频图像处理来实现动态交通流信息的采集技术能克服传统采集方式的缺陷,并已成为该研究领域的主流方向。本文所包含的内容如下:
作者在全面阅读和理解国内外智能交通系统研究领域的前沿技术的基础之上,研究了基于运动检测的交通流视频检测技术的原理。在分析当前各种技术的优缺点后,选择了基于双TMS320C6x DSP的交通流视频检测系统的设计方案,并选择了适合于嵌入式系统的算法—“背景差法”。针对二值化阈值自动优化选取问题,通过将模拟退火思想引入到遗传算法中设计了退火遗传算法(AGA),增强了整个系统的实时性和鲁棒性。同时结合“动态背景帧”、“视频触发”、“虚拟线圈”等方法,设计了车流量、车型、车速等交通参数的识别,并用VC++ 6.0开发了演示系统对算法进行了仿真。接着针对上位机开发了一个监控程序,总体上实现了四大功能:视频播放、串口通信、流量曲线显示、数据库查询。最后对算法的性能效果进行了分析,找出了问题原因并提出了未来改进的方向。
用本文的交通流视频检测算法对广州市和赣州市多段实况交通视频进行了反复地测试,取得了良好的效果。
With the rapid development of the modern economy, transportation has become the well known hot and difficult problem, we utilize the advanced technologies such as computer science, automation in the traffic of management and control to safeguard smooth and safe, it can also improve the quality of environment and promote the harmonious development of economy,then, Intelligent Traffic System turns up. Gathering the information of transportation timely is the key to the series of technologies for the intelligent traffic system. The technology of video detecting can overcome the flaw of traditional way, and has become the mainstream of this field. This paper contains the content as following:
After reading and understanding the advanced technology of the ITS, the author has studied the principle of video detection which is used in traffic flow based on the moving vehicle detection. After analysising the two sides of each kind of technology, this paper proposes the plan which is based on the double TMS320C6x DSP and it produces the algorithm of“the method of background frame difference”, which is fit for the embedded system. In view of the automatic selection of the best threshold which is used in the image division, here we introduce the thoughts of simulation of annealing into the basic genetic algorithm to generate the annealing genetic algorithm(AGA),then, the overall system’s timeliness and robustness are strengthened. Simultaneously , we have used a few of methods such as“dymatic background frame”,“hypothesized detection coil”,“Video triggering”and so on, to gather the traffic parameter such as traffic flow volume、vehicles’type and vehicles’speeds and so on. With VC++ 6.0, we have developed the demonstration system to realize the algorithm into reality.Then, the author developed a monitor software,which contains four modules:video player, Serial communication, volume curve display, database inquiry.Finally, it has carried on the analysis to the algorithm performance, then finding out the problems and the reasons ,it also point out the objection in future.
With the video detection algorithm, we have carried on video tests repeatedly under the environment of real road in Guangzhou and in Ganzhou, and the result is good .
作者在全面阅读和理解国内外智能交通系统研究领域的前沿技术的基础之上,研究了基于运动检测的交通流视频检测技术的原理。在分析当前各种技术的优缺点后,选择了基于双TMS320C6x DSP的交通流视频检测系统的设计方案,并选择了适合于嵌入式系统的算法—“背景差法”。针对二值化阈值自动优化选取问题,通过将模拟退火思想引入到遗传算法中设计了退火遗传算法(AGA),增强了整个系统的实时性和鲁棒性。同时结合“动态背景帧”、“视频触发”、“虚拟线圈”等方法,设计了车流量、车型、车速等交通参数的识别,并用VC++ 6.0开发了演示系统对算法进行了仿真。接着针对上位机开发了一个监控程序,总体上实现了四大功能:视频播放、串口通信、流量曲线显示、数据库查询。最后对算法的性能效果进行了分析,找出了问题原因并提出了未来改进的方向。
用本文的交通流视频检测算法对广州市和赣州市多段实况交通视频进行了反复地测试,取得了良好的效果。
With the rapid development of the modern economy, transportation has become the well known hot and difficult problem, we utilize the advanced technologies such as computer science, automation in the traffic of management and control to safeguard smooth and safe, it can also improve the quality of environment and promote the harmonious development of economy,then, Intelligent Traffic System turns up. Gathering the information of transportation timely is the key to the series of technologies for the intelligent traffic system. The technology of video detecting can overcome the flaw of traditional way, and has become the mainstream of this field. This paper contains the content as following:
After reading and understanding the advanced technology of the ITS, the author has studied the principle of video detection which is used in traffic flow based on the moving vehicle detection. After analysising the two sides of each kind of technology, this paper proposes the plan which is based on the double TMS320C6x DSP and it produces the algorithm of“the method of background frame difference”, which is fit for the embedded system. In view of the automatic selection of the best threshold which is used in the image division, here we introduce the thoughts of simulation of annealing into the basic genetic algorithm to generate the annealing genetic algorithm(AGA),then, the overall system’s timeliness and robustness are strengthened. Simultaneously , we have used a few of methods such as“dymatic background frame”,“hypothesized detection coil”,“Video triggering”and so on, to gather the traffic parameter such as traffic flow volume、vehicles’type and vehicles’speeds and so on. With VC++ 6.0, we have developed the demonstration system to realize the algorithm into reality.Then, the author developed a monitor software,which contains four modules:video player, Serial communication, volume curve display, database inquiry.Finally, it has carried on the analysis to the algorithm performance, then finding out the problems and the reasons ,it also point out the objection in future.
With the video detection algorithm, we have carried on video tests repeatedly under the environment of real road in Guangzhou and in Ganzhou, and the result is good .
引文
[1] http://jtj.dsq.gov.cn/1.htm
[2]普伟.基于视频的交通流量参数检测[D].西安电子科技大学硕士论文,2007
[3]张流.基于DSP的视频交通信息检测系统设计[D].天津大学硕士论文,2005
[4]钱亮.基于DMCU的交通流量检测系统研究与实现[D].东华大学硕士论文,2006
[5]董彩平.基于DSP和CPLD的运动目标跟踪系统的实现[D].天津大学硕士论文,2004
[6]魏星.智能视频监控系统中运动目标检测与跟踪技术的研究及其DSP实现[D].合肥工业大学硕士论文,2005
[7]杨威,张田文.复杂景物环境下运动目标检测的新方法[J].计算机研究与发展.1998,8(35)
[8] R. T. Collins, A. J. Lipton, T Kanade. A Systen for Vedio Surveillance and Monitoring. Proc.Am[C]. Nuclear Soc.(ANS) Eighth Int'1 Topical Meeting Robotic and Remote Systems, Apr. 1999
[9]王宾.视频序列中运动目标检测与跟踪有关问题的研究[D].西北大学硕士论文,2004
[10]四维科技,胡小锋等.Visual c++/MATLAB图像处理与识别使用案例精选,人民邮电大学出版社,2004.9
[11]张流.基于DSP的视频交通信息检测系统设计[D].天津大学硕士论文,2005
[12]许礼武.基于图像识别的交通数据采集终端设计[D].江西理工大学硕士论文2006
[13] Texas Instrument,TMS20C6211,TMS20C6211B Fixed-Point Digital Signal Processors
[14]陈永康.车辆视频检测算法研究[D].吉林大学硕士论文,2006
[15] C. R. Wren, A. Azarbaye jani, T. Darrel, Pfinder. real-time tracking of human body[J].IEEE Trans. Pattern Analysis and Machine Intelligence,780~785, 1997.7
[16] C. Stauffer and W. k. L. Grimson. Adaptive background mixture models for real-time tracking[C].Int. Conf. Computer Vision and Pattern Recognition, Vol. 2, 246~252, 1999
[17] A.Hlgammal,D.Harwood,and L.S.Davis.Non-parametric model for background subtraction[C],European Conf,Computer Vision,Vol. 2,751~767,2000
[18]任刚,崔霞.退火贪婪混合遗传算法[J].河南科学,2005,23(3):433-435
[19] Otsu, A threshold selection method from gray.1evel his-tograms [J] . IEEE Trans SMC,1979,9 (1): 62- 69
[20] http://www.chinaunix.net/jh/23/52955.html
[21]袁丁,基于视频和DSP的车辆识别测速系统,浙江大学硕士学位论文,2004
[22]赵祥模等.高速公路监控系统理论及应用,电子工业出版社,2003.11
[23] http://www.yesky.com/252/1837252_2.shtml
[24] http://www.gjwtech.com/vcandc/scommassistantcode.htm
[25]杨淑莹,边奠英.VC++图像处理程序设计,清华大学出版社,2003.11
[26]张雄伟,陈亮,徐光辉.DSP芯片的原理与开发应用,电子工业出版社,2003.2
[27]史忠科,曹力.交通图像检测与分析,科学出版社,2007.4
[28]彭仁明,贺春林,基于视频的车流量检测[J],西华师范大学学报(自然科学版),2004,12:404~407
[29]张晓东,动态交通流信息系统若干问题研究[D],吉林大学,2004,03
[30]李绪龙.基于图像处理技术的视频交通流信息采集系统研究[D].长安大学,2003,03
[31]魏武,基于计算机视觉和图像处理的交通参数检测[J],信息与控制,2001,06:257~260
[32]王耀南,李树涛,毛建旭.计算机图像处理与识别技术[M],高等教育出版社,2001
[33]张志勇,黄爱民,刘建平.视频智能交通系统[J],计算机工程与应用,2001,6:115~118
[34]王海虎,许志祥.智能交通检测与违章处理[J],电视技术,2000,11:77~80
[35]郁梅,蒋刚毅,郁伯康.智能交通系统中的计算机视觉技术应用[J],计算机工程,2001,10:101~104
[36]刘亚,艾海州,徐光佑.一种基于背景模型的运动目标检测与跟踪算法[J],信息与控,2002,08:315~319
[37]王夏黎,周明全,耿国华等.交通流视频检测系统的设计与实现[J],计算机应用,2004,09:68~71
[38]周长发,精通VC++图像编程[M],电子工业出版社,2000
[39]黄豪杰,李榕,常鸿森.基于自适应灰度形态学滤波的车牌图像分割[J].微电子学与计算机,2007,24(4).72-76
[40]赵金才,刘书桂,采用贪婪遗传算法实现图象阈值的自动选取[J].光电工程,2006,33(11):123-127
[41] Otsu N,A threshold selection method from gray.1evel his-tograms [J].IEEE Trans SMC, 1979,9 (1):62- 69
[42]华刚,郑南宁,薛建儒.基于改进遗传算法的边缘检测阈值自动选取及其应用[J].小型微型计算机系统,2002,23(3):318-321
[43]谢磊,朱光喜,张珍明.基于子特征的交通车辆检测及跟踪算法[J].微电子学与计算机,2007,24(7):65-69
[44]汤晖,许伦辉等.基于贪心退火遗传算法的车辆图像分割研究[J].交通与计算机,2007,25(4):19-22
[45]汤晖,李润,许伦辉.退火遗传算法的实现及其在交通检测中的应用[J].中国交通工程,2007,2:59-62
[46]邵春福,陈晓明等.基于视频图像处理的行人和非机动车数据采集技术[J]北京交通大学学报,2007,31(6):10-14
[2]普伟.基于视频的交通流量参数检测[D].西安电子科技大学硕士论文,2007
[3]张流.基于DSP的视频交通信息检测系统设计[D].天津大学硕士论文,2005
[4]钱亮.基于DMCU的交通流量检测系统研究与实现[D].东华大学硕士论文,2006
[5]董彩平.基于DSP和CPLD的运动目标跟踪系统的实现[D].天津大学硕士论文,2004
[6]魏星.智能视频监控系统中运动目标检测与跟踪技术的研究及其DSP实现[D].合肥工业大学硕士论文,2005
[7]杨威,张田文.复杂景物环境下运动目标检测的新方法[J].计算机研究与发展.1998,8(35)
[8] R. T. Collins, A. J. Lipton, T Kanade. A Systen for Vedio Surveillance and Monitoring. Proc.Am[C]. Nuclear Soc.(ANS) Eighth Int'1 Topical Meeting Robotic and Remote Systems, Apr. 1999
[9]王宾.视频序列中运动目标检测与跟踪有关问题的研究[D].西北大学硕士论文,2004
[10]四维科技,胡小锋等.Visual c++/MATLAB图像处理与识别使用案例精选,人民邮电大学出版社,2004.9
[11]张流.基于DSP的视频交通信息检测系统设计[D].天津大学硕士论文,2005
[12]许礼武.基于图像识别的交通数据采集终端设计[D].江西理工大学硕士论文2006
[13] Texas Instrument,TMS20C6211,TMS20C6211B Fixed-Point Digital Signal Processors
[14]陈永康.车辆视频检测算法研究[D].吉林大学硕士论文,2006
[15] C. R. Wren, A. Azarbaye jani, T. Darrel, Pfinder. real-time tracking of human body[J].IEEE Trans. Pattern Analysis and Machine Intelligence,780~785, 1997.7
[16] C. Stauffer and W. k. L. Grimson. Adaptive background mixture models for real-time tracking[C].Int. Conf. Computer Vision and Pattern Recognition, Vol. 2, 246~252, 1999
[17] A.Hlgammal,D.Harwood,and L.S.Davis.Non-parametric model for background subtraction[C],European Conf,Computer Vision,Vol. 2,751~767,2000
[18]任刚,崔霞.退火贪婪混合遗传算法[J].河南科学,2005,23(3):433-435
[19] Otsu, A threshold selection method from gray.1evel his-tograms [J] . IEEE Trans SMC,1979,9 (1): 62- 69
[20] http://www.chinaunix.net/jh/23/52955.html
[21]袁丁,基于视频和DSP的车辆识别测速系统,浙江大学硕士学位论文,2004
[22]赵祥模等.高速公路监控系统理论及应用,电子工业出版社,2003.11
[23] http://www.yesky.com/252/1837252_2.shtml
[24] http://www.gjwtech.com/vcandc/scommassistantcode.htm
[25]杨淑莹,边奠英.VC++图像处理程序设计,清华大学出版社,2003.11
[26]张雄伟,陈亮,徐光辉.DSP芯片的原理与开发应用,电子工业出版社,2003.2
[27]史忠科,曹力.交通图像检测与分析,科学出版社,2007.4
[28]彭仁明,贺春林,基于视频的车流量检测[J],西华师范大学学报(自然科学版),2004,12:404~407
[29]张晓东,动态交通流信息系统若干问题研究[D],吉林大学,2004,03
[30]李绪龙.基于图像处理技术的视频交通流信息采集系统研究[D].长安大学,2003,03
[31]魏武,基于计算机视觉和图像处理的交通参数检测[J],信息与控制,2001,06:257~260
[32]王耀南,李树涛,毛建旭.计算机图像处理与识别技术[M],高等教育出版社,2001
[33]张志勇,黄爱民,刘建平.视频智能交通系统[J],计算机工程与应用,2001,6:115~118
[34]王海虎,许志祥.智能交通检测与违章处理[J],电视技术,2000,11:77~80
[35]郁梅,蒋刚毅,郁伯康.智能交通系统中的计算机视觉技术应用[J],计算机工程,2001,10:101~104
[36]刘亚,艾海州,徐光佑.一种基于背景模型的运动目标检测与跟踪算法[J],信息与控,2002,08:315~319
[37]王夏黎,周明全,耿国华等.交通流视频检测系统的设计与实现[J],计算机应用,2004,09:68~71
[38]周长发,精通VC++图像编程[M],电子工业出版社,2000
[39]黄豪杰,李榕,常鸿森.基于自适应灰度形态学滤波的车牌图像分割[J].微电子学与计算机,2007,24(4).72-76
[40]赵金才,刘书桂,采用贪婪遗传算法实现图象阈值的自动选取[J].光电工程,2006,33(11):123-127
[41] Otsu N,A threshold selection method from gray.1evel his-tograms [J].IEEE Trans SMC, 1979,9 (1):62- 69
[42]华刚,郑南宁,薛建儒.基于改进遗传算法的边缘检测阈值自动选取及其应用[J].小型微型计算机系统,2002,23(3):318-321
[43]谢磊,朱光喜,张珍明.基于子特征的交通车辆检测及跟踪算法[J].微电子学与计算机,2007,24(7):65-69
[44]汤晖,许伦辉等.基于贪心退火遗传算法的车辆图像分割研究[J].交通与计算机,2007,25(4):19-22
[45]汤晖,李润,许伦辉.退火遗传算法的实现及其在交通检测中的应用[J].中国交通工程,2007,2:59-62
[46]邵春福,陈晓明等.基于视频图像处理的行人和非机动车数据采集技术[J]北京交通大学学报,2007,31(6):10-14