基于OpenCV的车辆视频检测技术研究
|
摘要
车辆检测是智能交通监控系统的关键环节。随着科技发展和技术进步,车辆检测方式正历经着由传统的线圈检测、雷达检测发展到更为先进的实时视频检测方式。车辆视频检测的基本原理是通过对从摄像机采集的视频序列通过数字图像处理技术,提取图像中的车辆信息,从而达到检测和判断车辆是否违章的目的。
视频检测是一个复杂的过程,不仅需要实时采集图像,而且很多图像处理算法不具有通用性,使得系统的开发周期延长,可扩展性不强。鉴于此,本文采用OpenCV作为开发平台,该库在Windows以及Linux下均可以使用,它是开放源代码并且完全免费,这就可以保证库的维护和更新更加便捷和迅速。OpenCV提供了很多基本图像处理方法的函数,并且征对视频图像处理特别提供了几种运动物体跟踪的原型,非常适合于车辆视频检测系统的开发。
论文设计并实现了一套基于OpenCV的视频车辆检测算法,算法综合考虑系统检测的可靠性和实时性。在算法流程中的背景初始化与更新步骤中,测试了现有的最常用的三种背景更新算法效果并总结其优劣,同时在此基础上提出了一种背景初始化的方法。在算法流程的另一个关键步骤阴影消除,提出了基于三种色彩空间RGB、YUV、HSV空间的阴影消除算法并进行了实验测试,阴影消除效果明显。在提取车辆特征上,采用状态机的方式对虚拟线圈区域判断当前是否有车辆经过,增强了检测系统的鲁棒性。
论文对所进行的一些工作进行了总结,并对车辆视频检测的发展进行了相关预测。
Vehicles detection is the key of an intelligent traffic monitoring system. With the development of technology, the methods of vehicles detection are developing from the traditional loop detection, radar detection to advanced real-time video detection. Its basic principle is that via digital image processing techniques extracting the vehicle information from the sequences of video image collected by cameras in order to achieve the purpose of detecting and determining whether the vehicle is illegal.
Not only Video detection requires collecting images in real time, but also many image processing algorithms are not universal, which makes the system development cycle to be delayed and the scalability of system to be not strong. In view of this, OpenCV is utilized as a development platform, which can be used under both Windows and Linux. It is an open source and completely free of charge so that the maintenance can be guaranteed and the update of its library can be more convenient and rapider. it provides lots of basic functions of image processing methods as well as several prototypes of tracking the moving objects aim at the video image processing.
An OpenCV-based video vehicle detecting algorithm is designed and achieved. In its background initialization and update processes, the existing three most common algorithms are tested to discuss their advantages and disadvantages, and then a new method of initializing the background is brought forward. In another key step of this algorithm, called shadow elimination, new shadow elimination algorithms based on three color space that are RGB, YUV and HSV space are brought forward. Also, we do experiments to show its much better effect on eliminating the shadow. In the features extraction of vehicles, the state machine is used on the virtual loop region to determine whether there are vehicles passing to enhance the robustness of the detection system.
The summary of our work are gave, and the development of video detection of vehicles is relevant forecasted.
视频检测是一个复杂的过程,不仅需要实时采集图像,而且很多图像处理算法不具有通用性,使得系统的开发周期延长,可扩展性不强。鉴于此,本文采用OpenCV作为开发平台,该库在Windows以及Linux下均可以使用,它是开放源代码并且完全免费,这就可以保证库的维护和更新更加便捷和迅速。OpenCV提供了很多基本图像处理方法的函数,并且征对视频图像处理特别提供了几种运动物体跟踪的原型,非常适合于车辆视频检测系统的开发。
论文设计并实现了一套基于OpenCV的视频车辆检测算法,算法综合考虑系统检测的可靠性和实时性。在算法流程中的背景初始化与更新步骤中,测试了现有的最常用的三种背景更新算法效果并总结其优劣,同时在此基础上提出了一种背景初始化的方法。在算法流程的另一个关键步骤阴影消除,提出了基于三种色彩空间RGB、YUV、HSV空间的阴影消除算法并进行了实验测试,阴影消除效果明显。在提取车辆特征上,采用状态机的方式对虚拟线圈区域判断当前是否有车辆经过,增强了检测系统的鲁棒性。
论文对所进行的一些工作进行了总结,并对车辆视频检测的发展进行了相关预测。
Vehicles detection is the key of an intelligent traffic monitoring system. With the development of technology, the methods of vehicles detection are developing from the traditional loop detection, radar detection to advanced real-time video detection. Its basic principle is that via digital image processing techniques extracting the vehicle information from the sequences of video image collected by cameras in order to achieve the purpose of detecting and determining whether the vehicle is illegal.
Not only Video detection requires collecting images in real time, but also many image processing algorithms are not universal, which makes the system development cycle to be delayed and the scalability of system to be not strong. In view of this, OpenCV is utilized as a development platform, which can be used under both Windows and Linux. It is an open source and completely free of charge so that the maintenance can be guaranteed and the update of its library can be more convenient and rapider. it provides lots of basic functions of image processing methods as well as several prototypes of tracking the moving objects aim at the video image processing.
An OpenCV-based video vehicle detecting algorithm is designed and achieved. In its background initialization and update processes, the existing three most common algorithms are tested to discuss their advantages and disadvantages, and then a new method of initializing the background is brought forward. In another key step of this algorithm, called shadow elimination, new shadow elimination algorithms based on three color space that are RGB, YUV and HSV space are brought forward. Also, we do experiments to show its much better effect on eliminating the shadow. In the features extraction of vehicles, the state machine is used on the virtual loop region to determine whether there are vehicles passing to enhance the robustness of the detection system.
The summary of our work are gave, and the development of video detection of vehicles is relevant forecasted.
引文
[1]王笑京,张纪升.中国智能交通系统(ITS)手册.第一版.北京:中国铁道出版社, 2008. 1~7
[2]李卫平.智能交通技术应用.第一版.北京:人民交通出版社, 2006. 1~10
[3]陈冰.美国智能交通系统的发展现状及规划.交通世界, 2002, 02(8): 34~38
[4]张扬,彭国雄,杨晓光.日本ITS发展现状及趋势.中外公路, 2003, 23(3): 8~10
[5] Kikuchi H., Kato K., Nakazato G. . Current Overview of ITS in JAPAN. Toward the Future 8th World Congress on ITS, 2005, 60(3): 264~280
[6]马颂德,张正友.计算机视觉-计算理论与算法基础.第一版.北京:科学出版社, 2003. 1~15
[7] Takabaetal S.. Measurement of Traffic Flow Using Real-Time Processing of Moving Pictures.in: 16th international Conference on industrial and Engineering Applications of Artificial interlligence and expert systems.UK: Springer Verlag, 2003. 488~494
[8] Abnameczuk T.. A Microcomputer Based TV Detector for Road Traffic.Symposium on Road Research Program, 1984, 1(10): 562~567
[9] Panos G. M.. Vehicles Detection through Image Processing. The Auto scope System. IEEE Transactions on Vehicular Technology, 1991, 40(1): 21~29
[10] Brian C.. Vision Makes Traffic Control Intelligent. Advanced Imaging, 1997, 12(2): 54~56
[11]王亮,胡卫明,谭铁牛.人运动的视觉分析综述.计算机学报, 2002, 25(3): 1~16
[12]佟守愚.基于视频技术的交通违章检测与识别理论及方法研究:[博士学位论文].吉林:吉林大学交通学院,2006
[13] Robert R., Goldberg L., Roth M.. Parallel algorithms for real time tracking. In: Proc. Of Machine Vision Applications, Architectures, and Systems Integration V. USA: Bellingham WA, 1996. 167~174
[14] Takatoo M.. Traffic flow measuring system using image processing. In: Proc of Automated Inspection and High-Speed Vision Architectures III. USA: BellinghamWA, 1989. 172~180
[15] Betke M.. Multiple vehicle detection and tracking. In: Proc of 25th AIPR Workshop: Emerging Applications of Computer Vision. USA: Bellingham WA, 1997. 104-110
[16]尹朝征.基于视频虚拟线圈的交通流参数检测:[硕士学位论文].北京:清华大学,2002
[17]陆其明. DirectShow实务精选.第一版.北京:北京科海电子出版, 2004. 45~62
[18]章毓晋.图像处理和分析技术.第二版.北京:高等教育出版社, 2008. 57~74
[19]黄爱民,安向京,骆力.数字图像处理与分析基础.第一版.北京:中国水利水电出版社, 2005. 123~150
[20] Rafael C., Gonzalez , Richard E. et al. Digital Image Processing. (Second Edition).阮秋琦,阮宇智.北京:电子工业出版社,2006. 117~170
[21]谢凤英,赵丹培. Visual C++数字图像处理.第二版.北京:电子工业出版社, 2006. 100~123
[22]汪国有.数字图像处理.第一版.武汉:华中科技大学出版社, 2009. 54~67
[23]李志慧,张长海,曲昭伟等.交通流视频检测中背景初始化算法.吉林大学学报. 2008, 38(01): 148~151
[24]卢强,陈泉林,林康红等.智能交通监视系统中路况图像背景去除算法研究.电子技术应用. 2001, 27(11): 35~36
[25]王亮生,程荫杭.一种改进的基于混合高斯分布模型的自适应背景消除算法.北京交通大学学报. 2003, 27(06): 22~25
[26]刘洁,张东来.关于自适应高斯混合背景模型的更新算法的研究.微计算机信息, 2008, 08(22): 241~243
[27]李全民,张运楚.自适应混合高斯背景模型的改进.计算机应用, 2007, 27(08): 2014~2017
[28]王典.基于混合高斯的背景建模与阴影抑制算法研究:[硕士学位论文] .西安:西北工业大学,2006
[29]胡小峰,赵辉.Vsiual C++/MATLAB图像处理与识别使用案例精选.第一版.北京:人民邮电出版社,2004. 85~92
[30]阴国富.基于阈值法的图像分割技术.现代电子技术, 2007, 30(23): 107~108
[31]阴躲芬,李一民,王英妹等.图像阈值分割技术的研究.科技广场,2008, 1(03): 136~137
[32] Otsu N.. A threshold selection method from gray-level histograms[J].IEEE Transactions on System Man and Cybernetic, 1979, 9(1): 62~66
[33] Jiang C., Ward M.O., Shadow detection for vechiles by locating the object-shadow boundary. In: Proc of Seventh IASTED International Conference on Signal and Image Processing, SIP 2005.Proc Seventh IASTED Int. Conf Sign Proc.UK: Acta Press. 2005. 315~319
[34]刘瑞祯,于仕琪.OpenCV教程基础篇.第一版.北京:北京航空航天大学出版社, 2007: 1~36
[35] Bradski G., Kaehler A.. Learning OpenCV(First Edition). USA: O’Reilly Media, 2008. 157~121
[36] Xiao M., Zhao C., Zhang H. L.. Moving Shadow Detection and Removal for Traffic Sequences. International Journal of Automation and Computing, 2007, 04(1): 38~46
[37]王军利,王树根.一种基于RGB彩色空间的影像阴影检测方法.信息技术, 2002, 1(12): 7~9
[38] Song K., Tai J.. Image-based Traffic Monitoring With Shadow Suppression . Proceedings of the IEEE, 2007, 95(2): 413~426
[39] Wang J.M., Chung Y.C., Chang C. L.. Shadow detection and removal for trafficimages. Networking, Sensing and Control, 2004 IEEE International Conference. 2004, 01(21): 649~654
[40]高岚,董慧颖,兰利宝等.自适应背景下运动目标阴影检测算法研究.现代电子技术,2007, 31(06): 59~61
[41]刘勃,魏铭旭,周荷琴等.交通场景中分块阴影检测算法研究.计算机工程, 2005, 31(11): 160~162, 219
[42]郭利生,郭立,焦荣惠等.一种基于运动阴影的目标检测算法.模式识别与人工智能,2007, 20(02): 180~184
[43]张丽,李志能.基于阴影检测的HSV空间自适应背景模型的车辆追踪检测.中国图像图形学报,2003, 08(07): 778~782
[44]钟琨,路小波,施毅等.一种基于YUV彩色空间的运动车辆阴影消除方法.交通与计算机,2007, 25(03): 110~113
[45] Wu Y. G., Tsai C. Y.. The Improvement of the Background Subtraction and Shadow Detection in Grayscale Video Sequences. In: Proc of International Machine Vision and Image Processing Conference, IMVIP 2007. Int. Mach Vis Image Process Conf. IMVIP. United States: Computer Society Press, 2007. 206
[46] Cezar J., Silveira J., Jung C. R.. Background Subtraction and Shadow Detection in Grayscale Video Sequences. In: Proc of 18th Brazilian Symposium on Computer Graphics and Image Processing. Braz. Symp. Comput. Graph.. United States: IEEE Computer Society, 2005. 189~196.
[47]王晓冬,霍宏,方涛等.基于快速归一化互相关函数的运动车辆阴影检测算法.计算机应用, 2006, 26(09): 2065~2067
[48]汪国有.图像与视频压缩技术讲义.第一版.武汉:华中科技大学图像识别与人工智能研究所, 2006, 22~27
[49]徐智穹,刘健,邱浩.有限状态机在嵌入式软件设计中的应用.智能电器及计算机应用, 2008, 06(13): 20~21, 28
[2]李卫平.智能交通技术应用.第一版.北京:人民交通出版社, 2006. 1~10
[3]陈冰.美国智能交通系统的发展现状及规划.交通世界, 2002, 02(8): 34~38
[4]张扬,彭国雄,杨晓光.日本ITS发展现状及趋势.中外公路, 2003, 23(3): 8~10
[5] Kikuchi H., Kato K., Nakazato G. . Current Overview of ITS in JAPAN. Toward the Future 8th World Congress on ITS, 2005, 60(3): 264~280
[6]马颂德,张正友.计算机视觉-计算理论与算法基础.第一版.北京:科学出版社, 2003. 1~15
[7] Takabaetal S.. Measurement of Traffic Flow Using Real-Time Processing of Moving Pictures.in: 16th international Conference on industrial and Engineering Applications of Artificial interlligence and expert systems.UK: Springer Verlag, 2003. 488~494
[8] Abnameczuk T.. A Microcomputer Based TV Detector for Road Traffic.Symposium on Road Research Program, 1984, 1(10): 562~567
[9] Panos G. M.. Vehicles Detection through Image Processing. The Auto scope System. IEEE Transactions on Vehicular Technology, 1991, 40(1): 21~29
[10] Brian C.. Vision Makes Traffic Control Intelligent. Advanced Imaging, 1997, 12(2): 54~56
[11]王亮,胡卫明,谭铁牛.人运动的视觉分析综述.计算机学报, 2002, 25(3): 1~16
[12]佟守愚.基于视频技术的交通违章检测与识别理论及方法研究:[博士学位论文].吉林:吉林大学交通学院,2006
[13] Robert R., Goldberg L., Roth M.. Parallel algorithms for real time tracking. In: Proc. Of Machine Vision Applications, Architectures, and Systems Integration V. USA: Bellingham WA, 1996. 167~174
[14] Takatoo M.. Traffic flow measuring system using image processing. In: Proc of Automated Inspection and High-Speed Vision Architectures III. USA: BellinghamWA, 1989. 172~180
[15] Betke M.. Multiple vehicle detection and tracking. In: Proc of 25th AIPR Workshop: Emerging Applications of Computer Vision. USA: Bellingham WA, 1997. 104-110
[16]尹朝征.基于视频虚拟线圈的交通流参数检测:[硕士学位论文].北京:清华大学,2002
[17]陆其明. DirectShow实务精选.第一版.北京:北京科海电子出版, 2004. 45~62
[18]章毓晋.图像处理和分析技术.第二版.北京:高等教育出版社, 2008. 57~74
[19]黄爱民,安向京,骆力.数字图像处理与分析基础.第一版.北京:中国水利水电出版社, 2005. 123~150
[20] Rafael C., Gonzalez , Richard E. et al. Digital Image Processing. (Second Edition).阮秋琦,阮宇智.北京:电子工业出版社,2006. 117~170
[21]谢凤英,赵丹培. Visual C++数字图像处理.第二版.北京:电子工业出版社, 2006. 100~123
[22]汪国有.数字图像处理.第一版.武汉:华中科技大学出版社, 2009. 54~67
[23]李志慧,张长海,曲昭伟等.交通流视频检测中背景初始化算法.吉林大学学报. 2008, 38(01): 148~151
[24]卢强,陈泉林,林康红等.智能交通监视系统中路况图像背景去除算法研究.电子技术应用. 2001, 27(11): 35~36
[25]王亮生,程荫杭.一种改进的基于混合高斯分布模型的自适应背景消除算法.北京交通大学学报. 2003, 27(06): 22~25
[26]刘洁,张东来.关于自适应高斯混合背景模型的更新算法的研究.微计算机信息, 2008, 08(22): 241~243
[27]李全民,张运楚.自适应混合高斯背景模型的改进.计算机应用, 2007, 27(08): 2014~2017
[28]王典.基于混合高斯的背景建模与阴影抑制算法研究:[硕士学位论文] .西安:西北工业大学,2006
[29]胡小峰,赵辉.Vsiual C++/MATLAB图像处理与识别使用案例精选.第一版.北京:人民邮电出版社,2004. 85~92
[30]阴国富.基于阈值法的图像分割技术.现代电子技术, 2007, 30(23): 107~108
[31]阴躲芬,李一民,王英妹等.图像阈值分割技术的研究.科技广场,2008, 1(03): 136~137
[32] Otsu N.. A threshold selection method from gray-level histograms[J].IEEE Transactions on System Man and Cybernetic, 1979, 9(1): 62~66
[33] Jiang C., Ward M.O., Shadow detection for vechiles by locating the object-shadow boundary. In: Proc of Seventh IASTED International Conference on Signal and Image Processing, SIP 2005.Proc Seventh IASTED Int. Conf Sign Proc.UK: Acta Press. 2005. 315~319
[34]刘瑞祯,于仕琪.OpenCV教程基础篇.第一版.北京:北京航空航天大学出版社, 2007: 1~36
[35] Bradski G., Kaehler A.. Learning OpenCV(First Edition). USA: O’Reilly Media, 2008. 157~121
[36] Xiao M., Zhao C., Zhang H. L.. Moving Shadow Detection and Removal for Traffic Sequences. International Journal of Automation and Computing, 2007, 04(1): 38~46
[37]王军利,王树根.一种基于RGB彩色空间的影像阴影检测方法.信息技术, 2002, 1(12): 7~9
[38] Song K., Tai J.. Image-based Traffic Monitoring With Shadow Suppression . Proceedings of the IEEE, 2007, 95(2): 413~426
[39] Wang J.M., Chung Y.C., Chang C. L.. Shadow detection and removal for trafficimages. Networking, Sensing and Control, 2004 IEEE International Conference. 2004, 01(21): 649~654
[40]高岚,董慧颖,兰利宝等.自适应背景下运动目标阴影检测算法研究.现代电子技术,2007, 31(06): 59~61
[41]刘勃,魏铭旭,周荷琴等.交通场景中分块阴影检测算法研究.计算机工程, 2005, 31(11): 160~162, 219
[42]郭利生,郭立,焦荣惠等.一种基于运动阴影的目标检测算法.模式识别与人工智能,2007, 20(02): 180~184
[43]张丽,李志能.基于阴影检测的HSV空间自适应背景模型的车辆追踪检测.中国图像图形学报,2003, 08(07): 778~782
[44]钟琨,路小波,施毅等.一种基于YUV彩色空间的运动车辆阴影消除方法.交通与计算机,2007, 25(03): 110~113
[45] Wu Y. G., Tsai C. Y.. The Improvement of the Background Subtraction and Shadow Detection in Grayscale Video Sequences. In: Proc of International Machine Vision and Image Processing Conference, IMVIP 2007. Int. Mach Vis Image Process Conf. IMVIP. United States: Computer Society Press, 2007. 206
[46] Cezar J., Silveira J., Jung C. R.. Background Subtraction and Shadow Detection in Grayscale Video Sequences. In: Proc of 18th Brazilian Symposium on Computer Graphics and Image Processing. Braz. Symp. Comput. Graph.. United States: IEEE Computer Society, 2005. 189~196.
[47]王晓冬,霍宏,方涛等.基于快速归一化互相关函数的运动车辆阴影检测算法.计算机应用, 2006, 26(09): 2065~2067
[48]汪国有.图像与视频压缩技术讲义.第一版.武汉:华中科技大学图像识别与人工智能研究所, 2006, 22~27
[49]徐智穹,刘健,邱浩.有限状态机在嵌入式软件设计中的应用.智能电器及计算机应用, 2008, 06(13): 20~21, 28