车道线检测在车道偏离预警中的应用研究
|
摘要
注意力分散是导致车道偏离而发生交通事故的常见原因之一。车道偏离预警系统实时探测车辆所在车道位置,若车辆偏离车道却没有开启相应的指示灯,系统将以光、声、振动等形式警告驾驶员,以保证行驶安全。由于路况的复杂性以及车辆行驶过程中要求系统具有较高的实时性,如何调节系统的鲁棒性、实时性和实用性以利于车道偏离预警系统能够投入到实际应用成为相关研究的重点。
在汽车辅助安全驾驶系统中,准确获取道路信息,即车道标志线的轨迹,是实现准确无误报警的前提和基础。本文对车道线检测技术的基本概念和基本理论进行分析和论述,在分析研究了国内外已有的车道线检测算法的基础上,对一些经典的车道线检测算法进行分析和研究,并根据LMedSquare曲线拟合具有减少噪声的特点,设计一种LMedSquare选取最佳车道线种子点集的直线拟合算法,并讨论了在不同道路路况时的解决方案。在道路路况较好的情况下,待拟合的种子点数目比较充足时,提出利用本文设计的一种选取最优待拟合种子点的最佳子集选择方案；在道路有遮挡、油污污染而导致道路车道线信息量少的路况下,根据图像帧之间具有连续性、相关性的特点,充分利用此前多帧车道线参数对当前图像的车道线参数进行卡尔曼预测,把符合风险函数取得最小值的参数最为作为当前车道线的参数。该算法在车道线检测过程中,有效解决了道路车道标识线遭到干扰而导致无法识别的问题。
本文研究、测试结果表明,基于LMedSquare选取最佳子集的车道线检测算法具有良好的检测效果,可应用到嵌入式集成视觉辅助行车安全系统实际项目中,实现辅助行车安全系统中的车道线参数的实时准确提取。
The Distraction of your mind which led to lane departure is one of the common causes of traffic accidents. Lane departure warning systems can detect the location of vehicles Real-timely, where if the vehicle deviated from the lane and does not open the corresponding indicator, the system will warning drivers by light, sound, vibration and other forms of warning. As the complexity of road conditions and the process of car running demands the system has high real-time, how to adjust the system robustness, and practicality in order to put lane departure warning system into practical application to be the focus of the research.
In the car safe driving system, accurate to obtain road information which is the path marking of the lane is the premise and foundation of accurate warning. In this thesis, the basic concepts and theories of lane detection technology was discussed.And analyzed some classics lane detection algorithm which has been studied abroad and home, and According to noise reduction of LMedSquare curve fitting, proposed that LMedSquare select the best set of seed points, and discussed the solutions of lane line detection algorithm in different road condition. In the favorable road condition, which caused the seed points are sufficient, design a options by which selected seed point of the best subset; when lane information is not clear which caused by barrier, oil pollution, in accordance with continuity and relativity successive the between image frames, we can full use of multi-frame line parameters which has been detected to predicted current lane parameters using Kalman filter, and than find the parameter which caused the risk function obtain the minimum value as the current lane parameters. This lane detection algorithm can detect lane parameter effective when lane has been disturbed.
The research and test shows that the testing lane algorithm based on LMedSquare select the best set of seed points has good effect, and can be applied to the actual project of embedded system visual display traffic safety system, to distill the lane parameters of traffic safety system accurately in real time.
在汽车辅助安全驾驶系统中,准确获取道路信息,即车道标志线的轨迹,是实现准确无误报警的前提和基础。本文对车道线检测技术的基本概念和基本理论进行分析和论述,在分析研究了国内外已有的车道线检测算法的基础上,对一些经典的车道线检测算法进行分析和研究,并根据LMedSquare曲线拟合具有减少噪声的特点,设计一种LMedSquare选取最佳车道线种子点集的直线拟合算法,并讨论了在不同道路路况时的解决方案。在道路路况较好的情况下,待拟合的种子点数目比较充足时,提出利用本文设计的一种选取最优待拟合种子点的最佳子集选择方案；在道路有遮挡、油污污染而导致道路车道线信息量少的路况下,根据图像帧之间具有连续性、相关性的特点,充分利用此前多帧车道线参数对当前图像的车道线参数进行卡尔曼预测,把符合风险函数取得最小值的参数最为作为当前车道线的参数。该算法在车道线检测过程中,有效解决了道路车道标识线遭到干扰而导致无法识别的问题。
本文研究、测试结果表明,基于LMedSquare选取最佳子集的车道线检测算法具有良好的检测效果,可应用到嵌入式集成视觉辅助行车安全系统实际项目中,实现辅助行车安全系统中的车道线参数的实时准确提取。
The Distraction of your mind which led to lane departure is one of the common causes of traffic accidents. Lane departure warning systems can detect the location of vehicles Real-timely, where if the vehicle deviated from the lane and does not open the corresponding indicator, the system will warning drivers by light, sound, vibration and other forms of warning. As the complexity of road conditions and the process of car running demands the system has high real-time, how to adjust the system robustness, and practicality in order to put lane departure warning system into practical application to be the focus of the research.
In the car safe driving system, accurate to obtain road information which is the path marking of the lane is the premise and foundation of accurate warning. In this thesis, the basic concepts and theories of lane detection technology was discussed.And analyzed some classics lane detection algorithm which has been studied abroad and home, and According to noise reduction of LMedSquare curve fitting, proposed that LMedSquare select the best set of seed points, and discussed the solutions of lane line detection algorithm in different road condition. In the favorable road condition, which caused the seed points are sufficient, design a options by which selected seed point of the best subset; when lane information is not clear which caused by barrier, oil pollution, in accordance with continuity and relativity successive the between image frames, we can full use of multi-frame line parameters which has been detected to predicted current lane parameters using Kalman filter, and than find the parameter which caused the risk function obtain the minimum value as the current lane parameters. This lane detection algorithm can detect lane parameter effective when lane has been disturbed.
The research and test shows that the testing lane algorithm based on LMedSquare select the best set of seed points has good effect, and can be applied to the actual project of embedded system visual display traffic safety system, to distill the lane parameters of traffic safety system accurately in real time.
引文
[1]董因平.高速汽车车道偏离预警系统的算法研究[D].长春：吉林大学汽车学院,2004
[2]毕雁冰.高速汽车车道偏离告警系统可行区域感知算法研究[D].长春：吉林大学汽车学院,2006
[3]毕雁冰,管欣,詹军.车道识别过程中搜索车道线的方法[J].汽车工程,2006,28(5)：339-442
[4]史忠科,曹力等.交通图像检测与分析[M].北京：科学出版社,2007.4：197-212.
[5]卢卫娜,史忠科.基于单目视觉的车道线和道路边界同步检测[J]传感技术学报2007.5
[6]吴沫.基于计算机视觉的车道偏离告警系统方法研究[D].长沙：国防科技大学,2005
[7]陆丰俊,基于视觉的车辆车道偏离预警系统算法研究[J],湖南长沙：华中科技大学,2006
[8]余天洪基于机器视觉的车道偏离预警系统研究[D]长春：吉林大学,2006
[9]葛平淑,车道偏离预警系统算法改进研究[D],长春：吉林大学,2008
[10]A Portable Real-Time Lane Departure Warning System based on Embedded Calculating Technique[J], Chang Gung University, Taiwan, IEEE2006.
[11]SukhanLee, WoongKwon, Jae-WonLee, A Vision Based Lane Departure Warning System[J], IEEE 2005
[12]Luo-Wei Tsai, Lane Detection Using Directional Random Walks[J], IEEE 2008.
[13]A.Lopez, C.Canero, J.Serrat Detection of Lane Markings based on Ridgeness and RANSAC[J]. IEEE Conference on Intelligent Transportation Systems 2005.10
[14]Axel Gern, Rainer Moebus, Vision-based Lane Recognition under Adverse Weather Conditions Using Optical Flow, IEEE,2002:652-657.
[15]易强,秦文虎.一种基于机器视觉的车道偏离报警系统[J].轻型汽车技术,2007,16-18
[16]刘洁,黄斌.单目视觉车道偏离报警系统[J].电子产品世界,2007.3,128-130
[17]赵佳佳.道路光照模式分类器设计[D].吉林：吉林大学交通学院,2007
[18]张伟.智能车辆中的道路检测与识别[D].重庆：重庆大学自动化学院,2006
[19]张云港.基于视觉的车道线检测算法[D].昆明：云南师范大学计算机科学与信息技术学院,2005
[20]陈军,赵玉凡,徐友春,彭永胜,袁一.基于启发式搜索的车道线识别算法研 究[J].计算机工程与应用,2007
[21]邓剑文.高速公路自主驾驶汽车道路检测技术研究[D].长沙：国防科学技术大学,2004
[22]曹国.汽车前方可行区域计算机视觉感知系统的研究[D]吉林大学2003
[23]邓自立.最优估计理论及其应用哈尔滨工业大学出版社[M].2005 108-186
[24]Batvaia, Dean A.Pomerieua, Predicting Lane Position for Roadway Departure Prevention, IEEE October,1998.
[25]J.W.Lee, CD.Kee, WK.Lee, A New Approach for Lane Departure Identification, IEEEIV2003:100-105.
[26]唐高友.基于机器视觉的道路识别与障碍物检测技术研究[D] 重庆大学2005
[27]Erez Dagan, Ofer Mano, Gideon, P.Stein, Amnon Shashua, Forward Collision Warning with a Single Camera. IEEE 2000:37-42.
[28]陈清阳.基于计算机视觉的车道跑偏告警系统研究[D]国防科技大学2007
[29]金钊.高速公路汽车车道偏离视觉导航系统的研究[D]武汉理工2007
[30]丁俊进.基于机器视觉道路识别技术的研究[D]武汉理工2007
[31]廖力.基于视觉的结构化道路检测算法研究[D]武汉理工2008
[32]龚建伟,王安帅,熊光明.一种自适应动态窗口车道线高速检测方法[J]北京理工大学学报2008.6
[33]刘涛,黄席樾.高速公路弯道识别算法[J]重庆大学学报2003
[34]陈清阳,李健,安向京.基于机器视觉的车道偏离预警系统的实现[J]嵌入式系统应用2008
[35]黄席樾,柴毅.汽车智能辅助操作系统中的障碍物检测[J]重庆大学学报2000
[36]ALLAN L.FISHER and PETER T.HIGHNAM Computing the Hough Transform on a Scan Line Array Processor, IEEE Transactions on pattern analysis and machine intelligence, VOl.11, No.3, MARCH 1989,262-265
[37]A. Broggi, S.Berte. Vision Based Road Detection in Automotive Systems[J] IEEE 2006
[38]郭磊,李克强.应用方向可调滤波器的车道线识别方法[J]机械工程学报2008.8
[39]M.Bertozzi and A.Broggi A Parallel Real Time Stereo Vision System for Generic Obstacle and Lane Detection[J], IEEE 1998:62-81.
[40]李若皓,丁冬花.一种基于扫描线的车道线识别算法[J]微计算机信息2008
[41]徐友春,王荣本.一种基于直线模型的道路识别算法研究[J]中国图像图形学报2004
[42]US NTSA,2002 Annual Assessment-Motor Vehicle Traffic Crash Fatality and Injury Estimates for 2002,2003.
[43]Yuji Kutsuma, Real-time Lane Line and Forward Vehicle Detection[J], ISCIT 2004
[44]王荣本,徐友春.基于线性模型的导航路径图像检测算法研究[J]公路交通科技2001
[45]唐佐林.结构化道路多行道线检测算法研究[J]南京理工大学2007
[46]ZHANG Z.Parameter estimation techniques:a tutorial with application to conic fitting, Image and Vision Computing Journal,1997:59-76
[47]金辉,吴乐林.结构化道路车道线识别的一种改进算法[J]北京理工大学学报2007.6
[48]郭磊,王建强,李克强.存在车辆干扰的车道线识别[J]汽车工程2007
[49]田炳香,郑榜贵.高速公路车道线检测与跟踪算法研究[J]智能交通2007
[50]董因平.高速汽车车道偏离预警系统研究[J]长春理工大学学报2004
[51]毕雁冰,管欣.基于稳态预瞄理论的一种车道偏离评价方法[J]中国机械工程2007.8
[52]毕雁冰.基于多边检测的一种车道偏离评价方法[J]深圳信息职业技术学院学报2007.5
[53]毕雁冰.提高车道线识别精度的方法[J]机电产品开发与创新2007.10
[54]朱彤,韩文涛,王波.改进hough变换方法在车道线检测中的应用[J]科研设计2004
[55]赵颖,王书茂,陈兵旗等.基于改进Hough变换的公路车道线快速检测算法[J].中国农业大学学报,2006
[56]侯德鑫,曹丽.一种基于视频图像的道路检测方法[J]仪器仪表学报2006.6
[2]毕雁冰.高速汽车车道偏离告警系统可行区域感知算法研究[D].长春：吉林大学汽车学院,2006
[3]毕雁冰,管欣,詹军.车道识别过程中搜索车道线的方法[J].汽车工程,2006,28(5)：339-442
[4]史忠科,曹力等.交通图像检测与分析[M].北京：科学出版社,2007.4：197-212.
[5]卢卫娜,史忠科.基于单目视觉的车道线和道路边界同步检测[J]传感技术学报2007.5
[6]吴沫.基于计算机视觉的车道偏离告警系统方法研究[D].长沙：国防科技大学,2005
[7]陆丰俊,基于视觉的车辆车道偏离预警系统算法研究[J],湖南长沙：华中科技大学,2006
[8]余天洪基于机器视觉的车道偏离预警系统研究[D]长春：吉林大学,2006
[9]葛平淑,车道偏离预警系统算法改进研究[D],长春：吉林大学,2008
[10]A Portable Real-Time Lane Departure Warning System based on Embedded Calculating Technique[J], Chang Gung University, Taiwan, IEEE2006.
[11]SukhanLee, WoongKwon, Jae-WonLee, A Vision Based Lane Departure Warning System[J], IEEE 2005
[12]Luo-Wei Tsai, Lane Detection Using Directional Random Walks[J], IEEE 2008.
[13]A.Lopez, C.Canero, J.Serrat Detection of Lane Markings based on Ridgeness and RANSAC[J]. IEEE Conference on Intelligent Transportation Systems 2005.10
[14]Axel Gern, Rainer Moebus, Vision-based Lane Recognition under Adverse Weather Conditions Using Optical Flow, IEEE,2002:652-657.
[15]易强,秦文虎.一种基于机器视觉的车道偏离报警系统[J].轻型汽车技术,2007,16-18
[16]刘洁,黄斌.单目视觉车道偏离报警系统[J].电子产品世界,2007.3,128-130
[17]赵佳佳.道路光照模式分类器设计[D].吉林：吉林大学交通学院,2007
[18]张伟.智能车辆中的道路检测与识别[D].重庆：重庆大学自动化学院,2006
[19]张云港.基于视觉的车道线检测算法[D].昆明：云南师范大学计算机科学与信息技术学院,2005
[20]陈军,赵玉凡,徐友春,彭永胜,袁一.基于启发式搜索的车道线识别算法研 究[J].计算机工程与应用,2007
[21]邓剑文.高速公路自主驾驶汽车道路检测技术研究[D].长沙：国防科学技术大学,2004
[22]曹国.汽车前方可行区域计算机视觉感知系统的研究[D]吉林大学2003
[23]邓自立.最优估计理论及其应用哈尔滨工业大学出版社[M].2005 108-186
[24]Batvaia, Dean A.Pomerieua, Predicting Lane Position for Roadway Departure Prevention, IEEE October,1998.
[25]J.W.Lee, CD.Kee, WK.Lee, A New Approach for Lane Departure Identification, IEEEIV2003:100-105.
[26]唐高友.基于机器视觉的道路识别与障碍物检测技术研究[D] 重庆大学2005
[27]Erez Dagan, Ofer Mano, Gideon, P.Stein, Amnon Shashua, Forward Collision Warning with a Single Camera. IEEE 2000:37-42.
[28]陈清阳.基于计算机视觉的车道跑偏告警系统研究[D]国防科技大学2007
[29]金钊.高速公路汽车车道偏离视觉导航系统的研究[D]武汉理工2007
[30]丁俊进.基于机器视觉道路识别技术的研究[D]武汉理工2007
[31]廖力.基于视觉的结构化道路检测算法研究[D]武汉理工2008
[32]龚建伟,王安帅,熊光明.一种自适应动态窗口车道线高速检测方法[J]北京理工大学学报2008.6
[33]刘涛,黄席樾.高速公路弯道识别算法[J]重庆大学学报2003
[34]陈清阳,李健,安向京.基于机器视觉的车道偏离预警系统的实现[J]嵌入式系统应用2008
[35]黄席樾,柴毅.汽车智能辅助操作系统中的障碍物检测[J]重庆大学学报2000
[36]ALLAN L.FISHER and PETER T.HIGHNAM Computing the Hough Transform on a Scan Line Array Processor, IEEE Transactions on pattern analysis and machine intelligence, VOl.11, No.3, MARCH 1989,262-265
[37]A. Broggi, S.Berte. Vision Based Road Detection in Automotive Systems[J] IEEE 2006
[38]郭磊,李克强.应用方向可调滤波器的车道线识别方法[J]机械工程学报2008.8
[39]M.Bertozzi and A.Broggi A Parallel Real Time Stereo Vision System for Generic Obstacle and Lane Detection[J], IEEE 1998:62-81.
[40]李若皓,丁冬花.一种基于扫描线的车道线识别算法[J]微计算机信息2008
[41]徐友春,王荣本.一种基于直线模型的道路识别算法研究[J]中国图像图形学报2004
[42]US NTSA,2002 Annual Assessment-Motor Vehicle Traffic Crash Fatality and Injury Estimates for 2002,2003.
[43]Yuji Kutsuma, Real-time Lane Line and Forward Vehicle Detection[J], ISCIT 2004
[44]王荣本,徐友春.基于线性模型的导航路径图像检测算法研究[J]公路交通科技2001
[45]唐佐林.结构化道路多行道线检测算法研究[J]南京理工大学2007
[46]ZHANG Z.Parameter estimation techniques:a tutorial with application to conic fitting, Image and Vision Computing Journal,1997:59-76
[47]金辉,吴乐林.结构化道路车道线识别的一种改进算法[J]北京理工大学学报2007.6
[48]郭磊,王建强,李克强.存在车辆干扰的车道线识别[J]汽车工程2007
[49]田炳香,郑榜贵.高速公路车道线检测与跟踪算法研究[J]智能交通2007
[50]董因平.高速汽车车道偏离预警系统研究[J]长春理工大学学报2004
[51]毕雁冰,管欣.基于稳态预瞄理论的一种车道偏离评价方法[J]中国机械工程2007.8
[52]毕雁冰.基于多边检测的一种车道偏离评价方法[J]深圳信息职业技术学院学报2007.5
[53]毕雁冰.提高车道线识别精度的方法[J]机电产品开发与创新2007.10
[54]朱彤,韩文涛,王波.改进hough变换方法在车道线检测中的应用[J]科研设计2004
[55]赵颖,王书茂,陈兵旗等.基于改进Hough变换的公路车道线快速检测算法[J].中国农业大学学报,2006
[56]侯德鑫,曹丽.一种基于视频图像的道路检测方法[J]仪器仪表学报2006.6