基于双目立体视觉的安全车距测量技术研究
|
摘要
双目立体视觉一个多学科交叉的领域,是计算机获取环境信息最重要的来源之一。近年来,图像科学的发展和计算机信号处理能力的增强为双目立体视觉的研究和应用提供了良好的条件,使双目立体视觉成为了计算机视觉领域的一个新的研究重点和热点。
双目立体视觉系统由于直接模拟了人类双眼处理景物的方式,且具有尺寸小、质量轻、性能可靠、信息量大、功耗小、噪声低、效率高、成本低、动态范围大以及光计量准确等优良特性,在许多领域极具应用价值。如自动驾驶、机器人导航与航测、工业检测、生产自动化、生物医学、航天遥测、军事侦察、三维测量学以及虚拟现实等。因此开展本课题的研究工作具有重要的理论意义、工程应用意义和实用价值。
本文在针对我国汽车保有量的迅速增加,公路上的交通事故发生率居高不下,交通安全问题日益突出的情况,以双目立体视觉技术、图像处理和模式识别等理论为基础,将车牌作为前方车辆的探测依据,提出了一种前方车辆的探测和安全车距测量方案。首先对双目立体视觉原理进行深入、系统地分析之后研究了两种不同的双目立体视觉系统,围绕双目立体视觉技术中的关键技术“双目立体视觉系统内外参数的高精度标定”,对各种标定模板及其控制点、摄像机模型的特点进行了比较和分析,结合应用环境确定了适合于标定的模板和控制点以及摄像机模型,并在此基础上研究了相应的摄像机标定方法,确定了标定方案。通过该方案标定得到的参数能够较好的校正存在畸变的原始图像、并适应车距测量要求。
针对双目立体视觉系统对目标车辆检测的需要,以车牌作车辆检测的依据,研究了基于灰度图像和基于彩色图像的车牌定位方法,设计了一种基于车牌号码区域灰度变化特征的定位汽车牌照方法,进行了相关实验。最后设计了试验环境并进行了实际车距测量,对结果进行了详细分析。试验结果表明了整个系统的高精度、大范围、可行性和有效性。
本文所做的工作为下一步双目立体视觉技术在障碍物避让和自动驾驶中路径规划的应用做了一些基础性工作,并为后续建立完善的基于双目立体视觉自动驾驶系统做了一些有益的尝试。
Binocular stereovision is a field of multi-subject crossed and one of the most important resources of computer accessing the environment information. Recently development of image science and enhancing of computer signal processing ability provide the well conditions for the study and application of binocular stereovision that becomes a new study emphasis and hotspot in computer vision field.
Binocular stereovision has great application value in many fields that include auto driving, robot navigation and aerial survey, industrial inspection, production automatization, biology medicine, space telemetry, military reconnaissance, three-dimensional measurement and virtual realization, because it directly simulates pattern of human eyes processing scenery and has choiceness characteristics such as little size, light mass, performances credibility, vast information quality, little power consumption, low noise, high efficiency, low cost, big dynamical range, light measure accuracy. So studying on binocular stereovision has important academic significance, engineering application significance and utility significance.
The thesis aims at rapidly increase of vehicle quantity in our nation, traffic accident occurring rate still in high position and traffic security problems increasingly outstanding, based on theories of binocular stereovision, image processing and pattern recognition, and puts forward a project of exploration and measuring safe distance of frontage vehicles according license plate as exploration of frontage vehicles. Firstly profound systemically studied on theory of binocular stereovision and two different kinds of binocular stereovision system, round the key technology of inside and outside parameters high accuracy pointing of binocular stereovision system in binocular stereovision, compare and analyze the many kinds of calibration pattern and control point and camera model. This thesis, combined the application environment, defines fit demarcating pattern and control point and camera model, and analyzes methods of camera demarcating based on those, confirms the demarcating project. Parameters that make from the project can emendate fault image and adapt the demands of measuring vehicle distance.
In allusion to needs of binocular stereovision system measuring object vehicle and according vehicle brand as measuring, this thesis studies orientation method of vehicle brand based on gray image and color image, and designs a kind of orientation method of vehicle brand, based on vehicle brand area gray changing characteristic. Finally designs the experiment environment and takes the experiments which results been analyzed on detail. Results of these experiments indicate that the system is high precision, big range, feasibility and validity.
This thesis has done a lot of prophase work of auto driving system with binocular stereovision. The work not only is the basis of binocular stereovision application of roadblock avoiding and route layout in auto driving system, but also makes the basis of consummate auto driving system with binocular stereovision.
双目立体视觉系统由于直接模拟了人类双眼处理景物的方式,且具有尺寸小、质量轻、性能可靠、信息量大、功耗小、噪声低、效率高、成本低、动态范围大以及光计量准确等优良特性,在许多领域极具应用价值。如自动驾驶、机器人导航与航测、工业检测、生产自动化、生物医学、航天遥测、军事侦察、三维测量学以及虚拟现实等。因此开展本课题的研究工作具有重要的理论意义、工程应用意义和实用价值。
本文在针对我国汽车保有量的迅速增加,公路上的交通事故发生率居高不下,交通安全问题日益突出的情况,以双目立体视觉技术、图像处理和模式识别等理论为基础,将车牌作为前方车辆的探测依据,提出了一种前方车辆的探测和安全车距测量方案。首先对双目立体视觉原理进行深入、系统地分析之后研究了两种不同的双目立体视觉系统,围绕双目立体视觉技术中的关键技术“双目立体视觉系统内外参数的高精度标定”,对各种标定模板及其控制点、摄像机模型的特点进行了比较和分析,结合应用环境确定了适合于标定的模板和控制点以及摄像机模型,并在此基础上研究了相应的摄像机标定方法,确定了标定方案。通过该方案标定得到的参数能够较好的校正存在畸变的原始图像、并适应车距测量要求。
针对双目立体视觉系统对目标车辆检测的需要,以车牌作车辆检测的依据,研究了基于灰度图像和基于彩色图像的车牌定位方法,设计了一种基于车牌号码区域灰度变化特征的定位汽车牌照方法,进行了相关实验。最后设计了试验环境并进行了实际车距测量,对结果进行了详细分析。试验结果表明了整个系统的高精度、大范围、可行性和有效性。
本文所做的工作为下一步双目立体视觉技术在障碍物避让和自动驾驶中路径规划的应用做了一些基础性工作,并为后续建立完善的基于双目立体视觉自动驾驶系统做了一些有益的尝试。
Binocular stereovision is a field of multi-subject crossed and one of the most important resources of computer accessing the environment information. Recently development of image science and enhancing of computer signal processing ability provide the well conditions for the study and application of binocular stereovision that becomes a new study emphasis and hotspot in computer vision field.
Binocular stereovision has great application value in many fields that include auto driving, robot navigation and aerial survey, industrial inspection, production automatization, biology medicine, space telemetry, military reconnaissance, three-dimensional measurement and virtual realization, because it directly simulates pattern of human eyes processing scenery and has choiceness characteristics such as little size, light mass, performances credibility, vast information quality, little power consumption, low noise, high efficiency, low cost, big dynamical range, light measure accuracy. So studying on binocular stereovision has important academic significance, engineering application significance and utility significance.
The thesis aims at rapidly increase of vehicle quantity in our nation, traffic accident occurring rate still in high position and traffic security problems increasingly outstanding, based on theories of binocular stereovision, image processing and pattern recognition, and puts forward a project of exploration and measuring safe distance of frontage vehicles according license plate as exploration of frontage vehicles. Firstly profound systemically studied on theory of binocular stereovision and two different kinds of binocular stereovision system, round the key technology of inside and outside parameters high accuracy pointing of binocular stereovision system in binocular stereovision, compare and analyze the many kinds of calibration pattern and control point and camera model. This thesis, combined the application environment, defines fit demarcating pattern and control point and camera model, and analyzes methods of camera demarcating based on those, confirms the demarcating project. Parameters that make from the project can emendate fault image and adapt the demands of measuring vehicle distance.
In allusion to needs of binocular stereovision system measuring object vehicle and according vehicle brand as measuring, this thesis studies orientation method of vehicle brand based on gray image and color image, and designs a kind of orientation method of vehicle brand, based on vehicle brand area gray changing characteristic. Finally designs the experiment environment and takes the experiments which results been analyzed on detail. Results of these experiments indicate that the system is high precision, big range, feasibility and validity.
This thesis has done a lot of prophase work of auto driving system with binocular stereovision. The work not only is the basis of binocular stereovision application of roadblock avoiding and route layout in auto driving system, but also makes the basis of consummate auto driving system with binocular stereovision.
引文
[1]荣德.美国公路交通安全的新焦点.海外视窗,2005,8:75-77
[2]郑文贵,李向云.1996-2003年全国交通事故伤害的时间序列分析.卫生事业管理,2006,2:105-107
[3]扬子江,韩永强,宋敏群.2004年全国道路交通事故情况分析报告.道路交通管理,2005,1:5-7
[4]庄继德.汽车地面运输系统工程.北京理工大学出版社,1999
[5]林芝.智能汽车的距离测量技术.汽车与安全,2002,3:44-47
[6]C.Thorpe.Vision and navigation:The Carnegie Mellon Navlab.Kluwer Academic Publishers,1999:52-72
[7]T.Jochem.A portable navigation platform.IEEE Symposium on Intelligent,Detroit,Michigan,1995:107-112
[8]W.Huber.Autonoumous Driving on Vehicle Test Tracks:Overview Motivation,and Concept Proc.IEEE Conference Intelligent Vehicle,1998:445-450
[9]A.Broggi.Automatic Vehicle Guidance:The Experience of the ARGO Autonomous Vehicle.World Science Publishing Co.Ltd,1999:25-32
[10]刘刚,候德藻等.汽车纵向行驶安装报警系统开发.汽车工程,2004,26(3):291-294
[11]游素亚,徐光拓.立体视觉研究的现状与进展.中国图象图形学报,1997,2(1):17-24
[12]岁波,都东,陈强,等.基于双目视觉的工业机器人运动轨迹准确度检测.机械工程学报,2003,39(5):88-91
[13]高庆吉,洪炳熔,阮玉峰.基于异构双目视觉的全自主足球机器人导航.哈尔滨工业大学学报,2003,35(9):1029-1032,1088
[14]Wang J.Li Y 3D object modeling using a binocular vision system.Proc.IEEE International Conference on Instrumentation and Measurement Technology,Venice,Italy,1999:684-99
[15]Bernardino A,Santos J.Visual behaviours for binocular tracking.Robotics and Autonomous Systems,1998,25:137-146
[16]巴拉德 D,布朗 C.计算机视觉.北京:科学出版社,1987
[17]高文,陈熙霖.计算机视觉—算法与系统原理.北京:清华大学出版社,1999
[18]马颂德,张正友.计算机视觉—计算理论与算法基础.北京:科学出版,1998
[19]Roberts L.Machine perception of three-dimensional solids.Optical and Electro-Optical Information Processing,MIT Press,Cambridge,1965,159-197
[20]Renzhao Jin,Zhaoyi Wang.Introduction to Virtual Reality and Applications in Agriculture.Journal of Tianjin Agricultural College,2001,8(2):27-32
[21]王润生.图像理解.长沙:国防科技大学出版社,1995
[22]Y.Ruichek,H.Issa,J.G.Postaire et al.Towards real-time obstacle detection using a hierarchical decomposition methodology for stereo matching with a genetic algorithm.16th IEEE International Conference,2004:138-147
[23]章毓晋.图像理解,北京:清华大学出版社,2007.2
[24]Faig W.Calibration of close-range photogrammetry systems:mathematical formulation.Photogrammetric Eng.Remote Sensing,1975,41(12):1479-1486
[25]Abdel Y,Karara H.Direct linear transformation into object space coordinates in close-range photogrammetry.Proceedings of the Symposium on Close-Range Photogrammetry,1971:1-18
[26]Luh J,Klaasen J.A three dimensional vsion by off-shelf system with multi-cameras.IEEE Trans.PAMI,1985,7:195-210
[27]Ganapathy S.Decomposition of transformation matrices for robot vision.Pattern Recognition Letters,1989,2:410-412
[28]Tsai R Y.A versatile camera calibration technique for high-accuracy 3D machine vision metrology using off-the-shelf TV camera and lenses[J].IEEE Journal of Robotics and Automation,1987,3(4):323-344
[29]Zhang,Z.A flexible new technique for camera calibration.IEEE Transactions on Pattern Analysis and Machine Intelligence,1996,26(10):44-48
[30]Faugeras O,Maybank S.Motion from point matches:Multiplicity of solutions.Int.Journal of Computer Vision,1990,4:255-246
[31]Maybank S,Faugeras D.A theory of self-calibration of a moving camera.International Journal of Computer Vision,1992,8(2):123-151
[32]MA S.A self-calibration technique for active vision.IEEE Trans.on Robotics and Automation,1996,12(1):114-120
[33]吴文琪,孙增圻.机器视觉中的摄像机定标方法综述.计算机应用研究,2004,21(2):4-6
[34]马尔.计算机视觉(姚国正,刘磊,汪云九,译者).北京:科学出版社,1988
[35]刘伟铭,赵雪平.一种基于扫描行的汽车车牌定位算法.计算机工程与应用,2004,06:223-226
[36]卢雅琴,凌超.于数学形态学的车牌定位方法.计算机工程,2005,03:224-226
[37]杨俊,戚飞虎.一种基于形状和纹理特征的车牌定位方法.计算机工程,2006,02:199-202
[38]金一粟,袁宝民,于万波,等.基于分形盒子维数的车牌定位方法.计算机应用研究,2002.09:40-44
[39]余棉水,黎绍发.基于边缘与SVM的车牌自动定位与提取.计算机应用研究,2004,10:131-133
[40]张玉庆.智能交通系统中的车牌定位问题研究.哈尔滨:哈尔滨工程大学,2003.12
[41]许礼武,许伦辉,黄艳国.基于小波分解的车牌定位算法.计算机工程,2006,21:191-196
[42]熊军,高敦堂,都思丹,等.应用遗传算法进行车牌定位.计算机应用,2004,06:163-167
[43]郭火波.彩色汽车图像车牌定位技术分析.山西大学学报,2005,01:40-43
[44]柳作民,刘彦戎,沈言行,等.GB.12676-1999,汽车制动系统结构、性能和试验方法.1999:9-12
[2]郑文贵,李向云.1996-2003年全国交通事故伤害的时间序列分析.卫生事业管理,2006,2:105-107
[3]扬子江,韩永强,宋敏群.2004年全国道路交通事故情况分析报告.道路交通管理,2005,1:5-7
[4]庄继德.汽车地面运输系统工程.北京理工大学出版社,1999
[5]林芝.智能汽车的距离测量技术.汽车与安全,2002,3:44-47
[6]C.Thorpe.Vision and navigation:The Carnegie Mellon Navlab.Kluwer Academic Publishers,1999:52-72
[7]T.Jochem.A portable navigation platform.IEEE Symposium on Intelligent,Detroit,Michigan,1995:107-112
[8]W.Huber.Autonoumous Driving on Vehicle Test Tracks:Overview Motivation,and Concept Proc.IEEE Conference Intelligent Vehicle,1998:445-450
[9]A.Broggi.Automatic Vehicle Guidance:The Experience of the ARGO Autonomous Vehicle.World Science Publishing Co.Ltd,1999:25-32
[10]刘刚,候德藻等.汽车纵向行驶安装报警系统开发.汽车工程,2004,26(3):291-294
[11]游素亚,徐光拓.立体视觉研究的现状与进展.中国图象图形学报,1997,2(1):17-24
[12]岁波,都东,陈强,等.基于双目视觉的工业机器人运动轨迹准确度检测.机械工程学报,2003,39(5):88-91
[13]高庆吉,洪炳熔,阮玉峰.基于异构双目视觉的全自主足球机器人导航.哈尔滨工业大学学报,2003,35(9):1029-1032,1088
[14]Wang J.Li Y 3D object modeling using a binocular vision system.Proc.IEEE International Conference on Instrumentation and Measurement Technology,Venice,Italy,1999:684-99
[15]Bernardino A,Santos J.Visual behaviours for binocular tracking.Robotics and Autonomous Systems,1998,25:137-146
[16]巴拉德 D,布朗 C.计算机视觉.北京:科学出版社,1987
[17]高文,陈熙霖.计算机视觉—算法与系统原理.北京:清华大学出版社,1999
[18]马颂德,张正友.计算机视觉—计算理论与算法基础.北京:科学出版,1998
[19]Roberts L.Machine perception of three-dimensional solids.Optical and Electro-Optical Information Processing,MIT Press,Cambridge,1965,159-197
[20]Renzhao Jin,Zhaoyi Wang.Introduction to Virtual Reality and Applications in Agriculture.Journal of Tianjin Agricultural College,2001,8(2):27-32
[21]王润生.图像理解.长沙:国防科技大学出版社,1995
[22]Y.Ruichek,H.Issa,J.G.Postaire et al.Towards real-time obstacle detection using a hierarchical decomposition methodology for stereo matching with a genetic algorithm.16th IEEE International Conference,2004:138-147
[23]章毓晋.图像理解,北京:清华大学出版社,2007.2
[24]Faig W.Calibration of close-range photogrammetry systems:mathematical formulation.Photogrammetric Eng.Remote Sensing,1975,41(12):1479-1486
[25]Abdel Y,Karara H.Direct linear transformation into object space coordinates in close-range photogrammetry.Proceedings of the Symposium on Close-Range Photogrammetry,1971:1-18
[26]Luh J,Klaasen J.A three dimensional vsion by off-shelf system with multi-cameras.IEEE Trans.PAMI,1985,7:195-210
[27]Ganapathy S.Decomposition of transformation matrices for robot vision.Pattern Recognition Letters,1989,2:410-412
[28]Tsai R Y.A versatile camera calibration technique for high-accuracy 3D machine vision metrology using off-the-shelf TV camera and lenses[J].IEEE Journal of Robotics and Automation,1987,3(4):323-344
[29]Zhang,Z.A flexible new technique for camera calibration.IEEE Transactions on Pattern Analysis and Machine Intelligence,1996,26(10):44-48
[30]Faugeras O,Maybank S.Motion from point matches:Multiplicity of solutions.Int.Journal of Computer Vision,1990,4:255-246
[31]Maybank S,Faugeras D.A theory of self-calibration of a moving camera.International Journal of Computer Vision,1992,8(2):123-151
[32]MA S.A self-calibration technique for active vision.IEEE Trans.on Robotics and Automation,1996,12(1):114-120
[33]吴文琪,孙增圻.机器视觉中的摄像机定标方法综述.计算机应用研究,2004,21(2):4-6
[34]马尔.计算机视觉(姚国正,刘磊,汪云九,译者).北京:科学出版社,1988
[35]刘伟铭,赵雪平.一种基于扫描行的汽车车牌定位算法.计算机工程与应用,2004,06:223-226
[36]卢雅琴,凌超.于数学形态学的车牌定位方法.计算机工程,2005,03:224-226
[37]杨俊,戚飞虎.一种基于形状和纹理特征的车牌定位方法.计算机工程,2006,02:199-202
[38]金一粟,袁宝民,于万波,等.基于分形盒子维数的车牌定位方法.计算机应用研究,2002.09:40-44
[39]余棉水,黎绍发.基于边缘与SVM的车牌自动定位与提取.计算机应用研究,2004,10:131-133
[40]张玉庆.智能交通系统中的车牌定位问题研究.哈尔滨:哈尔滨工程大学,2003.12
[41]许礼武,许伦辉,黄艳国.基于小波分解的车牌定位算法.计算机工程,2006,21:191-196
[42]熊军,高敦堂,都思丹,等.应用遗传算法进行车牌定位.计算机应用,2004,06:163-167
[43]郭火波.彩色汽车图像车牌定位技术分析.山西大学学报,2005,01:40-43
[44]柳作民,刘彦戎,沈言行,等.GB.12676-1999,汽车制动系统结构、性能和试验方法.1999:9-12
© 2004-2018 中国地质图书馆版权所有 京ICP备05064691号 京公网安备11010802017129号 地址：北京市海淀区学院路29号 邮编：100083 电话：办公室：(+86 10)66554848；文献借阅、咨询服务、科技查新：66554700 |