基于车道检测的车辆检测方法
|
摘要
智能交通系统(ITS)是未来交通系统的发展方向,它是将先进的信息技术、数据通讯传输技术、电子传感技术、控制技术及计算机技术等有效地集成运用于整个地面交通管理系统而建立的一种在大范围内、全方位发挥作用的、实时、准确、高效的综合交通运输管理系统。先进的交通检测技术和计算处理技术是智能交通系统中的一个重要研究领域。
本文以交通路口单个固定摄像头拍摄的视频图像为研究对象,提出了一种基于车道检测的车辆检测方法,将通常运用于车载系统的车道线检测技术运用于环境较复杂的交通路口车辆检测系统,为车辆检测自动划分感兴趣区域,大大降低了车辆检测的计算量。
论文首先总结了常用的目标检测方法及背景建模方法,并对它们进行分析和对比,从而选择背景帧差法进行车辆检测,选择时间中值法进行背景建模,并且提出了自己的背景更新策略。
随后,本文根据所得到的背景图像的特征,对图像进行滤波处理。在对常用的图像预处理算法进行总结和对比后,选择了适合道路背景图像的边缘检测算法,提出了一种基于灰度值选择的道路背景滤波方法,并且对传统的Hough变换方法进行了改进。
然后,本文根据得到的车道信息,划分了车辆检测的感兴趣区域,使用阈值分割法将车辆目标分割出来,采用连通域标识法得到目标的最小外接矩形,通过计算外接矩形的个数来完成车辆计数。
最后,本文使用固定相机拍摄的一段2分47秒的路面视频来对算法进行检验。实验表明,本文提出的算法可以有效地检测出晴好天气下结构化道路上的车道线,对车辆的检测有较高的正确率。尽管如此,本文还有许多有待改进的地方,如还需要进一步的研究,使得本系统能检测更多的交通车辆参数,比如对车辆速度、长度等的检测;也需要进一步的改进,使得该系统能运用于环境更为复杂的非结构化道路中。
Intelligent Transportation System (ITS) is the development trend of transportation system. Advanced information technology, data communication Technology, electronic sensing technology, control technology and computer technology are integrated in ITS, which constitute the comprehensive, effective, accurate and real-time transportation system. Advanced traffic monitoring and computer processing technology is an important area in ITS.
The videos captured by a static camera are studied in this paper, and a vehicle detection method based on road detection is proposed. The road detection which is typically used in in-car system, is used to determine the interest areas of vehicle detection. It efficiently reduces the complexity of the vehicle detection.
The paper first gives a summary of motion detection methods and background modeling methods, analyzes and compares them with each other, and chooses the background subtraction method to detect the vehicles, uses Time Median Background Model Method to form a background image, then gives a background update method.
In the second part, it uses filters to reduce the noises, according to the features of the background image. The summary and comparison of image pre-processing algorithm is discussed in this paper, and the appropriate edge detection method is chosen, a road detection algorithm based on gray value chosen is raised, the traditional Hough Transform is improved.
According to the information of roads, it determines the interest areas of vehicle detection, uses a threshold operation to get the moving object, and apply Clustering algorithm to form the MERs(Minimum Enclosing Rectangle), and then takes count of MERs to finish the vehicle counted.
At the end, a road video which continues 2minute and 47second , is used to test the system improved by this paper. Test Results show our system can efficiently detect the road-line in clear day, and get a high rate of recognition. However, it still requires more work on how to compute traffic parameters in the future, like speed, vehicle length, etc. It also need more research on how to apply this system to the unstructured road which have a more flexible environment.
本文以交通路口单个固定摄像头拍摄的视频图像为研究对象,提出了一种基于车道检测的车辆检测方法,将通常运用于车载系统的车道线检测技术运用于环境较复杂的交通路口车辆检测系统,为车辆检测自动划分感兴趣区域,大大降低了车辆检测的计算量。
论文首先总结了常用的目标检测方法及背景建模方法,并对它们进行分析和对比,从而选择背景帧差法进行车辆检测,选择时间中值法进行背景建模,并且提出了自己的背景更新策略。
随后,本文根据所得到的背景图像的特征,对图像进行滤波处理。在对常用的图像预处理算法进行总结和对比后,选择了适合道路背景图像的边缘检测算法,提出了一种基于灰度值选择的道路背景滤波方法,并且对传统的Hough变换方法进行了改进。
然后,本文根据得到的车道信息,划分了车辆检测的感兴趣区域,使用阈值分割法将车辆目标分割出来,采用连通域标识法得到目标的最小外接矩形,通过计算外接矩形的个数来完成车辆计数。
最后,本文使用固定相机拍摄的一段2分47秒的路面视频来对算法进行检验。实验表明,本文提出的算法可以有效地检测出晴好天气下结构化道路上的车道线,对车辆的检测有较高的正确率。尽管如此,本文还有许多有待改进的地方,如还需要进一步的研究,使得本系统能检测更多的交通车辆参数,比如对车辆速度、长度等的检测;也需要进一步的改进,使得该系统能运用于环境更为复杂的非结构化道路中。
Intelligent Transportation System (ITS) is the development trend of transportation system. Advanced information technology, data communication Technology, electronic sensing technology, control technology and computer technology are integrated in ITS, which constitute the comprehensive, effective, accurate and real-time transportation system. Advanced traffic monitoring and computer processing technology is an important area in ITS.
The videos captured by a static camera are studied in this paper, and a vehicle detection method based on road detection is proposed. The road detection which is typically used in in-car system, is used to determine the interest areas of vehicle detection. It efficiently reduces the complexity of the vehicle detection.
The paper first gives a summary of motion detection methods and background modeling methods, analyzes and compares them with each other, and chooses the background subtraction method to detect the vehicles, uses Time Median Background Model Method to form a background image, then gives a background update method.
In the second part, it uses filters to reduce the noises, according to the features of the background image. The summary and comparison of image pre-processing algorithm is discussed in this paper, and the appropriate edge detection method is chosen, a road detection algorithm based on gray value chosen is raised, the traditional Hough Transform is improved.
According to the information of roads, it determines the interest areas of vehicle detection, uses a threshold operation to get the moving object, and apply Clustering algorithm to form the MERs(Minimum Enclosing Rectangle), and then takes count of MERs to finish the vehicle counted.
At the end, a road video which continues 2minute and 47second , is used to test the system improved by this paper. Test Results show our system can efficiently detect the road-line in clear day, and get a high rate of recognition. However, it still requires more work on how to compute traffic parameters in the future, like speed, vehicle length, etc. It also need more research on how to apply this system to the unstructured road which have a more flexible environment.
引文
[1]宣颖.智能交通系统体系及关键技术研究.[硕士学位论文].国防科技大学 2003:10-12
[2]冯奇,周小鹏,孙立军,沈祺.视频图像处理技术在实时交通信息采集中的应用.上海公路,2004(1):50-53
[3]卢湖川,朱矿军.基于视频的交通流参数检测.城市交通,2005(02):70-74
[4]杨明,宋雪峰,王宏,张钹.面向智能交通系统的图像处理.计算机工程与应用,华北计算技术研究所:2001(9):4-8
[5]李文举.智能交通中图像处理技术应用的研究.大连:大连海事大学,2005
[6]宋正勋,胡贞,许红梅.DSP器件的原理及应用.长春光学精密机械学院学报,2003,16(2):149-152
[7]王永生.基于DSP平台的视频处理器的设计与应用.[硕士学位论文].陕西:西北工业大学,2006
[8]李纪奎,向怀坤,曹泉,刘建伟.面向视频应用的DSP内存管理技术研究.交通与计算机,2005,(5):94-97
[9]王仁龙,闫保中等.基于DM642下视频采集压缩系统的技术研究.弹箭与制导学报(自然科学版),2007,(1):235-237
[10]刘文.高速数字信号采集处理系统设计.应用科技,2003,30(7):27-32
[11]陈宏,杨兆选.交通信息视频检测系统设计.电子测量技术,2006,29(2):139-140
[12]王伟,李志鹏,刘允才.交通信息实时视频检测系统的实现.微型电脑应用,2006,22(9):26-28
[13]郁梅,蒋刚毅,贺赛龙.基于路面标记的车辆检测和计数[J].仪器仪表学报,2002,(8):387-390
[14]隋晔,马钺.交通监控系统中运动目标分类和跟踪研究.信息与控制,2003,32(1):61-64
[15]朱军,叶庆泰.基于DSP的运动目标识别与跟踪.机械,Machinery,2003,30(1):12-17
[16]卢强,陈泉林,奉玲,林康红.视频交通监控系统的开发.计算机应用,2001(12):69-70
[17]贺春林.一种基于视频的车辆检测算法.计算机科学,2005(5):243-245
[18]林洪文,涂丹,李国辉.基于统计背景模型的目标检测方法.计算机工程,2003(9),29(16):97-99
[19]章毓晋.图像处理与分析.北京:清华大学出版社,1999.181-184
[20]何斌,马天予,王运坚等.Visual C++数字图像处理(第二版).北京:人民邮电出版社,2002.395-400
[21]Mei Yu,Yong Deak Kim.Vertical edge based algorithm for Korean license plate recognition,The Journal of The Korean Institute of Communication Sciences,2000,25(7A):1076-1083
[22]C.Harris M.Stephens.A combined corner and edge detector[M].In fourth Alvey Vision Conf.,1988.147-151.
[23]Manfred Kopp and Werner Purgathofer,"efficient 3x3 median filter",Available:http://citeseer.ist.psu.edu/67259.html.
[24]冯星奎,肖兴明.方向加权中值滤波算法.中国图像图形学报.2000.4
[25]刘志敏,杨杰.基于数学形态学的图像形态滤波.红外与激光工程.1999.4
[26]许礼武,许伦辉,黄艳国,刘文亮.基于DSP交通数据采集系统的设计.交通与计算机,2006(1):107-109
[27]叶海炳.基于视频的汽车流量检测.重庆大学,2004
[28]Masayoshi Aoki.Image Processing in ITS[J].IEEE International Conference on Intelligent Vehicles,1998:1-4
[29]Bertozzi M,Broggi.AVision-based Vehicle Guidance[J].IEEE Computer,1997,30(7):227-234
[30]Alan J.LIpyon,Hironobu Fujiyoushi,Rajus.Patil.Moving Target Classification and Tracking from Real-time Video.IEEE Intelligent Systems November,1998
[31]Ohta,N.A statistical approach to background subtraction for surveillance systems.Computer Vision,2000.ICCV 2001.Proceedings,Eighth IEEE International Conference,2001,2(2):481-486
[32]Ying Ren,Chin-Seng Chua,Yeong-Khing Ho.Motion detection with non- stationary background Image Analysis and Processing,2001.Proceedings.11th International Conference,2001:78-83
[33]王怀理,王德生,田立生.Hough变换OCR图像倾斜矫正方法.中国图像图形学报,2001.02
[34]韩秋蕾,朱明,姚志军.基于改进Hough变换的图像线段特征提取.仪器仪表学报.2004(8),25(4),436-439
[35]康传利,姚连壁.利用Hough变换和稳健估计计算道路线形参数.公路工程.2007,32(6),7-9
[36]Lurng-Kuo Liu,Feig,E.A block-based gradient descent search algorithm for block motion estimation in video coding.IEEE Transactions on Circuits and Systems for Video Technology,1996,6(4):419-422
[37]Moghaddam B,Nastar C,Pentland A.A Bayesian similarity measure for direct image matching[C].In..Proc int'l Conf Pattern Recognition,1996:350-357
[38]Yajun Fang,Marcelo,Mizuki,Ichiro Masaki and Berthoud Horn.TV Camera-Based Vehicle Detection and its Chip Implementation.IEEE Intelligent Vehicles Symposium 2000Dearborn(MI),USA October 3-5,2000
[39]Osama Masoud,Nikolaos P,Papanikolopoulos Robust Pedestrian Tracking using A Model-Based Approach,IEEE Intelligent systems June,1998
[40]Takatoo.M.et.Traffic flow measuring system using image processing.SPIEm,1999,1197:172-180
[41]袁佑云,牛兴和,刘传云.基于最小外接矩形的稻米粒型检测方法.粮食与饲料工业. 2006, 9: 7-10
[42] Yao Nie, Kai-Kuang Ma. Adaptive rood pattern search for fast block- matching motion estimation. IEEE Transactions on Image Processing, 2002, 11(12): 1442-1449
[43] Zhigang Zhu, Guangyou Xu, BoYang, etal. VISATRAM: areal-time vision system for automatic traffic monitoring. Trans[J].Image and Vision Computing, 2000 (18): 781-794.
[2]冯奇,周小鹏,孙立军,沈祺.视频图像处理技术在实时交通信息采集中的应用.上海公路,2004(1):50-53
[3]卢湖川,朱矿军.基于视频的交通流参数检测.城市交通,2005(02):70-74
[4]杨明,宋雪峰,王宏,张钹.面向智能交通系统的图像处理.计算机工程与应用,华北计算技术研究所:2001(9):4-8
[5]李文举.智能交通中图像处理技术应用的研究.大连:大连海事大学,2005
[6]宋正勋,胡贞,许红梅.DSP器件的原理及应用.长春光学精密机械学院学报,2003,16(2):149-152
[7]王永生.基于DSP平台的视频处理器的设计与应用.[硕士学位论文].陕西:西北工业大学,2006
[8]李纪奎,向怀坤,曹泉,刘建伟.面向视频应用的DSP内存管理技术研究.交通与计算机,2005,(5):94-97
[9]王仁龙,闫保中等.基于DM642下视频采集压缩系统的技术研究.弹箭与制导学报(自然科学版),2007,(1):235-237
[10]刘文.高速数字信号采集处理系统设计.应用科技,2003,30(7):27-32
[11]陈宏,杨兆选.交通信息视频检测系统设计.电子测量技术,2006,29(2):139-140
[12]王伟,李志鹏,刘允才.交通信息实时视频检测系统的实现.微型电脑应用,2006,22(9):26-28
[13]郁梅,蒋刚毅,贺赛龙.基于路面标记的车辆检测和计数[J].仪器仪表学报,2002,(8):387-390
[14]隋晔,马钺.交通监控系统中运动目标分类和跟踪研究.信息与控制,2003,32(1):61-64
[15]朱军,叶庆泰.基于DSP的运动目标识别与跟踪.机械,Machinery,2003,30(1):12-17
[16]卢强,陈泉林,奉玲,林康红.视频交通监控系统的开发.计算机应用,2001(12):69-70
[17]贺春林.一种基于视频的车辆检测算法.计算机科学,2005(5):243-245
[18]林洪文,涂丹,李国辉.基于统计背景模型的目标检测方法.计算机工程,2003(9),29(16):97-99
[19]章毓晋.图像处理与分析.北京:清华大学出版社,1999.181-184
[20]何斌,马天予,王运坚等.Visual C++数字图像处理(第二版).北京:人民邮电出版社,2002.395-400
[21]Mei Yu,Yong Deak Kim.Vertical edge based algorithm for Korean license plate recognition,The Journal of The Korean Institute of Communication Sciences,2000,25(7A):1076-1083
[22]C.Harris M.Stephens.A combined corner and edge detector[M].In fourth Alvey Vision Conf.,1988.147-151.
[23]Manfred Kopp and Werner Purgathofer,"efficient 3x3 median filter",Available:http://citeseer.ist.psu.edu/67259.html.
[24]冯星奎,肖兴明.方向加权中值滤波算法.中国图像图形学报.2000.4
[25]刘志敏,杨杰.基于数学形态学的图像形态滤波.红外与激光工程.1999.4
[26]许礼武,许伦辉,黄艳国,刘文亮.基于DSP交通数据采集系统的设计.交通与计算机,2006(1):107-109
[27]叶海炳.基于视频的汽车流量检测.重庆大学,2004
[28]Masayoshi Aoki.Image Processing in ITS[J].IEEE International Conference on Intelligent Vehicles,1998:1-4
[29]Bertozzi M,Broggi.AVision-based Vehicle Guidance[J].IEEE Computer,1997,30(7):227-234
[30]Alan J.LIpyon,Hironobu Fujiyoushi,Rajus.Patil.Moving Target Classification and Tracking from Real-time Video.IEEE Intelligent Systems November,1998
[31]Ohta,N.A statistical approach to background subtraction for surveillance systems.Computer Vision,2000.ICCV 2001.Proceedings,Eighth IEEE International Conference,2001,2(2):481-486
[32]Ying Ren,Chin-Seng Chua,Yeong-Khing Ho.Motion detection with non- stationary background Image Analysis and Processing,2001.Proceedings.11th International Conference,2001:78-83
[33]王怀理,王德生,田立生.Hough变换OCR图像倾斜矫正方法.中国图像图形学报,2001.02
[34]韩秋蕾,朱明,姚志军.基于改进Hough变换的图像线段特征提取.仪器仪表学报.2004(8),25(4),436-439
[35]康传利,姚连壁.利用Hough变换和稳健估计计算道路线形参数.公路工程.2007,32(6),7-9
[36]Lurng-Kuo Liu,Feig,E.A block-based gradient descent search algorithm for block motion estimation in video coding.IEEE Transactions on Circuits and Systems for Video Technology,1996,6(4):419-422
[37]Moghaddam B,Nastar C,Pentland A.A Bayesian similarity measure for direct image matching[C].In..Proc int'l Conf Pattern Recognition,1996:350-357
[38]Yajun Fang,Marcelo,Mizuki,Ichiro Masaki and Berthoud Horn.TV Camera-Based Vehicle Detection and its Chip Implementation.IEEE Intelligent Vehicles Symposium 2000Dearborn(MI),USA October 3-5,2000
[39]Osama Masoud,Nikolaos P,Papanikolopoulos Robust Pedestrian Tracking using A Model-Based Approach,IEEE Intelligent systems June,1998
[40]Takatoo.M.et.Traffic flow measuring system using image processing.SPIEm,1999,1197:172-180
[41]袁佑云,牛兴和,刘传云.基于最小外接矩形的稻米粒型检测方法.粮食与饲料工业. 2006, 9: 7-10
[42] Yao Nie, Kai-Kuang Ma. Adaptive rood pattern search for fast block- matching motion estimation. IEEE Transactions on Image Processing, 2002, 11(12): 1442-1449
[43] Zhigang Zhu, Guangyou Xu, BoYang, etal. VISATRAM: areal-time vision system for automatic traffic monitoring. Trans[J].Image and Vision Computing, 2000 (18): 781-794.