车牌识别系统中定位算法的研究
|
摘要
随着信息技术和智能技术的发展,交通管理系统的信息化、智能化已成为发展的趋势。车牌识别系统(LPR)是智能交通系统(ITS)的核心组成部分,在现代交通收费管理系统中发挥着举足轻重的作用。近年来,对车牌识别系统中关键技术的研究已经成为智能交通领域的一个热点问题。本文对车牌识别系统中的图像预处理、车牌定位和字符分割等几大关键部分进行了比较深入、全面的论述,并对关键定位技术进行了深入的研究。
本文采用基于颜色信息的综合车牌定位方法,车牌定位通过以下几个步骤来实现:首先是车牌粗定位,根据车牌区域横向积分投影连续性的特点,同时利用车牌白点数目占据主导的特点,用一个比估计的车牌小的矩形遍历整个边缘二值图,提取出大致的车牌范围。然后进行车牌底色的判断,因为在车牌图像预处理时已经保存了车牌颜色方面的信息,在此时分析出车牌的底色有利于下一步的车牌精确定位;其次是精确定位,根据车牌颜色的像素占该候选车牌区域所有像素的比例,采用行方向和列方向上的车牌定位技术,由此得到比较精确的车牌区域;然后采用结合Hough变换和垂直投影的方法对车牌进行倾斜校正,接着对字符图像进行去除边界处理,最后对字符图像进行切分。
通过对采集于各种真实环境的图像进行实验,结果表明,本文所采用的方法能达到较好的车牌定位和字符分割效果,具有一定的鲁棒性和实时性。
With the development of information technology and intelligence technology, the informatization and intelligentizing of traffic management is the trend. License Plate Recognition system (LPR) is the core of Intelligent Traffic System (ITS). It is very important in modern traffic management systems. In recent years, the study on the critical techniques for LPR has already become an important research field of scientific circles. In this paper we discuss the three parts of License Plate Recognition Systems: image pre-process, license plate location and character segmentation, and make a deep research on some important and key technology.
A hybrid license plate location scheme is presented in this paper, which is mainly implemented as following several stages. At first, in the license plate roughly location stage, according to the continuity of transverse integral projection of the license plate region, and according to the number of white change points taking up a majority of the license plate, we use a rectangle to traverse the whole marginal-binary-image of the plate rage. After extracting the approximate license plate, judgment of the base color of the license plate is done, because the information of color is conserved when we do pre-treatment of the images. Second stage is accurately extracting the license plate. We use a method including horizontal and vertical operators to locate the license plate region; the pixel-rate is utilized to verify the real license plates which are the pixel of the plate color compare to the pixel of the candidate license plate color. In the last section, the slant correction of license plate is necessary to use the method combined with Hough transform and projection, so that the correspond plate region is present. Then we excise the border of the characters and segment them.
Based on the experiments dealing with images taken under various real world conditions, the results prove that this proposed method can relatively locate license plate and segment characters. It shows that the performance of the system is promising, robust and timely.
本文采用基于颜色信息的综合车牌定位方法,车牌定位通过以下几个步骤来实现:首先是车牌粗定位,根据车牌区域横向积分投影连续性的特点,同时利用车牌白点数目占据主导的特点,用一个比估计的车牌小的矩形遍历整个边缘二值图,提取出大致的车牌范围。然后进行车牌底色的判断,因为在车牌图像预处理时已经保存了车牌颜色方面的信息,在此时分析出车牌的底色有利于下一步的车牌精确定位;其次是精确定位,根据车牌颜色的像素占该候选车牌区域所有像素的比例,采用行方向和列方向上的车牌定位技术,由此得到比较精确的车牌区域;然后采用结合Hough变换和垂直投影的方法对车牌进行倾斜校正,接着对字符图像进行去除边界处理,最后对字符图像进行切分。
通过对采集于各种真实环境的图像进行实验,结果表明,本文所采用的方法能达到较好的车牌定位和字符分割效果,具有一定的鲁棒性和实时性。
With the development of information technology and intelligence technology, the informatization and intelligentizing of traffic management is the trend. License Plate Recognition system (LPR) is the core of Intelligent Traffic System (ITS). It is very important in modern traffic management systems. In recent years, the study on the critical techniques for LPR has already become an important research field of scientific circles. In this paper we discuss the three parts of License Plate Recognition Systems: image pre-process, license plate location and character segmentation, and make a deep research on some important and key technology.
A hybrid license plate location scheme is presented in this paper, which is mainly implemented as following several stages. At first, in the license plate roughly location stage, according to the continuity of transverse integral projection of the license plate region, and according to the number of white change points taking up a majority of the license plate, we use a rectangle to traverse the whole marginal-binary-image of the plate rage. After extracting the approximate license plate, judgment of the base color of the license plate is done, because the information of color is conserved when we do pre-treatment of the images. Second stage is accurately extracting the license plate. We use a method including horizontal and vertical operators to locate the license plate region; the pixel-rate is utilized to verify the real license plates which are the pixel of the plate color compare to the pixel of the candidate license plate color. In the last section, the slant correction of license plate is necessary to use the method combined with Hough transform and projection, so that the correspond plate region is present. Then we excise the border of the characters and segment them.
Based on the experiments dealing with images taken under various real world conditions, the results prove that this proposed method can relatively locate license plate and segment characters. It shows that the performance of the system is promising, robust and timely.
引文
[1] 王笑京,ITS 在中国的发展,http://www.itsc.com.cn.
[2] 蔡文沁,我国智能交通系统发展的战略构想.交通运输系统工程与信息,2003 年 2月,Vol 3, No.l:16-22.
[3] Javier Cano and Juan-Carlos P′erez-Cort′es F.J. Perales et al. (Eds.): Vehicle License Plate Segmentation in Natural Images, IbPRIA 2003, LNCS 2652, 2003._c Springer-Verlag Berlin Heidelberg 2003:142-149.
[4] 尹令,车牌自动识别系统的研究及其在 VC++中的实现,[硕士学位论文],湘潭, 湘潭大学计算控制技术与控制理论应用,2003:1-3.
[5] Deploying the Integrated Metropolitan ITS Infrastructure FY2000 Report,FHWA,U. S. Department of Transportation , 2001.
[6] FHWA FTA, NHTSA, FMCSA. 2002 Intelligent Transportation Systems(ITS) ProjectsBook. HTML Version. U. S. Department of Transportation ITS Joint Program Office,2002.
[7] 陈旭梅,于雷,郭继孚等,美国智能交通系统 ITS 的近期发展综述,中外公路,2003 Vo1.23, No.2: 9-120.
[8] 刘馨月,复杂背景下车牌分割技术的研究,[硕士学位论文],吉林长春,东北师范大学,2006:2-5.
[9] Barroso J,Bulas-Cruz J & Dagless E L,Real-Time Number Plate Reading[J]. 4th IFAC Workshop on Algorithms and Architectures for Real-Time Control,1997.
[10] R.Parisi. Car Plate Recognition by Neural Networks and Image Processing[J]. IEEE International Symposium on Industrial Electronics, 1998:605-610.
[11] C.H.Coetzee,et al. PC Based Number Plate Recognition Systems, In Proc.IEEE International Symposium on Industrial Electronics. 1999,1(1):605-610.
[12] C.Rittscher, J.Kato, S.Joga, and A.Blake., A Probabilistic Background Model for Tracking. 2000, 2(2): 63-67.
[13] 刘效静,成瑜.,汽车牌照自动识别技术研究,[J].南京航天航空大学学报, 1998,30(5):573-575.
[14] 牛欣,沈兰荪,基于特征的车辆牌照定位算法,[J].交通与计算机,. 2001,1(81):31-33.
[15] 刘智勇,刘迎建,车牌识别(LPR)中的图像提取及分割,[J].中文信息学报,2000,14(4):29-33.
[16] 郁梅,郁伯康,郑义,基于视觉的车辆牌照检测,[J].计算机应用研究. 1999(5):65-67.
[17] 廖金周,宣国荣,车辆牌照的自动分割,[J]. 微型电脑应用,1999(7):32-34.
[18] 戚飞虎,叶芗芸,李卫东等,机动车辆自动收费系统,[J].交通与计算机,1997,15(1):38-41.
[19] 戴青云,余英林,一种基于小波与形态学的车牌图像分割方法,[J].中国图形图像学报,2000,5A(5):411-415.
[20] 万国红,王敏,黄心汉等,基于神经网络的汽车牌照自动识别技术研究,[J]. 计算机工程与应用,2002,(6):204-205.
[21] 赵雪春,戚飞虎,基于彩色分割的车牌自动识别技术,[J].上海交通大学学报,1998,(10):3-9.
[22] 张引 , 潘云鹤,彩色汽车图像牌照定位新方法, [J]. 中国图像图形学报 .2001,6(4):374-377.
[23] 何斌,马天予,王运坚等,Visual C++数字图像处理,[M].北京:人民邮电出版社,2002.
[24] David J.Kruglinski, Scot Wingo, George Shepherd 著 , 希 望 图 书 创 作 室译.ProgrammingVisual C++6.0 技术内幕,[M].北京,北京希望电子出版社,2001.
[25] 苏明金,刘宏,刘波,Matlab 高级编程,电子工业出版社,2005:45-79.
[26] 陈永超,李汉强,Matlab7.0 与 VC6.0 的接口技术,电脑知识与技术,2006.1:129-130, SSN 1009-3044.
[27] 王琴,车辆牌照提取与识别算法的研究,[硕士学位论文],武汉,武汉理工大学,2005:11-15.
[28] 朱志刚,数字图像处理基础,[M].北京,清华大学出版社,1996.
[29] 王强,马利庄,图像二值化时图像特征的保留,计算机辅助设计与图形学学报,Vol.12,No.10,Oct,2000.
[30] N.Ostu,A threshold selection method from gay level histogram, EEE Trans. Sys.Man,Cybern,9(1),1979:62-66.
[31] 王润生,图像理解,国防科技大学出版社,1995
[32] J.Bernsen,Dvnamic thresholdingof gay-level images,Proc.of 8th Int'l Conf. onPatt. Recongn,1986:1251-1255.
[33] Byeong Rae Lee, Kyungsoo Park, Hyunchul Kang ,et al. Adaptive Local Binarization Method for Recognition of Vehicle License Plates,IWCIA 2004, LNCS 3322, Springer-Verlag Berlin Heidelberg 2004:646–655.
[34] 张铁弓,基于 DSP 的微光视频图像实时处理研究,[硕士学位论文],武汉,武汉理工大学,2002:33-36.
[35] 胡小锋,赵辉,《Visual C++/MATLAB 图像处理与识别实用案例精选》,[M]. 北京,人民邮电出版社,2004.
[36] Marr D, HILDR ETH E, Theory of edge detection[J]. Proc.Royal Soc. London, B,1980,(207):187-217.
[37] 付仲良,黄书强,陈眠等.货车图像车牌区快速定位及字符切割算法.计算机工 程与设计,2003, Vol.24 No.l:77-79.
[38] T Randen, J.H Husoy. Filtering for texture classification: a comparative study, IEEE Trans.Pattern Analysis and Machine Intelligence, 1999, Vo1.21 No.4.
[39] 陈昕,基于纹理分析的车牌图像区域分割技术的研究,[硕士学位论文]合肥,合肥工业大学 2003:17-32.
[40] 汤德佑,黄元江,基于灰度跳变的车牌自动识别技术,株洲工学院学报,Mar. 2004 Vol. 18 No. 2: 1-2.
[41] 熊军,高敦堂,沈庆宏等,基于字符纹理特征的快速定位算法,光电工程,2003, Vol. 30 No. 2: 60-63.
[42] 林纯青,徐立亚,戚飞虎,汽车图象中字符目标的提取算法,上海交通大学学报,1998, Vol.33 No. 10:1-3.
[43] 张西宁,郑南宁,汽车牌照自动识别中的目标与背景的快速分割,信息与控制,1998: 27-31.
[44] 金玲玲,廖芹,汪刘一,汽车牌照的提取方法,华南理工大学学报,2002,Vo1.30 No.7:95-97.
[45] 杨海廷,基于纹理特征的车牌识别系统的研究与实现,[硕士学位论文],成都,电子科技大学,2005:22-27.
[46] Barroso J, Rafael A. Number Plate Reading Using Computer Vision[J]. IEEE Transaction on Vehicular Technology. 2001.11.
[47] Rittscher C, Kato J. A Probabilistic Background Model for Tracking. Eccv 2000,Vol.2.
[48] 宋 建 才 , 字 符 结 构 知 识 在 车 牌 识 别 中 的 应 用 , 电 子 技 术 应 用 ,2004,Vol.30,No.5 :188-191.
[49] 郭大波,陈礼民,卢朝阳等,基于车牌底色识别的车牌定位方法,计算机工程与设计,2004,Vol.24.No.124,No5:81-84.
[50] 王建平,姜滔,基于小波分析的汽车牌照分割,合肥工业大学学报,2002 o1.25 No.6:1139-1142.
[51] Parist R. Car Plate Recognition by Neural Networks and Image Processing IEEE International Symposium on Circuits and Systems, USA, 2000, May 31-June 3.
[52] 崔江,王友仁,车牌自动识别方法中的关键技术研究,[J]. 计算机测量与控制,2003,Vol.11 No.4:260-262.
[53] 刘彦婷,徐海兰,杨磊,车牌识别软件的改进算法,北京广播学院学报(自然科学版), 2005年12月第12卷,第4期:39-43.
[54] 芮挺,沈春林,张金林.车牌识别中倾斜牌照的快速矫正算法,[J].计算机工程 2004,30(13):122-124.
[55] 王华,丁晓青,一种多字体印刷藏文字符的归一化方法,[J].计算机应用研究,2004,(6):41-43.
[56] David Llorens, Andres Marzal, et al. Car License Plates Extraction and Recognition Based on Connected Components Analysis and HMM Decoding[J]. IbPRIA 2005,LNCS 3522,Springer-Verlag Berlin Heidelberg,2005:571–578.
[57] 张云刚,张长水,利用 Hough 变换和先验知识的车牌字符分割算法,[J].计算机学报,2004.1:130-135.
[58] 陈寅鹏,丁晓青,复杂车辆图像中的车牌定位与字符分割方法,[J]. 红外与激光工程,2004,33(1):29-33.
[2] 蔡文沁,我国智能交通系统发展的战略构想.交通运输系统工程与信息,2003 年 2月,Vol 3, No.l:16-22.
[3] Javier Cano and Juan-Carlos P′erez-Cort′es F.J. Perales et al. (Eds.): Vehicle License Plate Segmentation in Natural Images, IbPRIA 2003, LNCS 2652, 2003._c Springer-Verlag Berlin Heidelberg 2003:142-149.
[4] 尹令,车牌自动识别系统的研究及其在 VC++中的实现,[硕士学位论文],湘潭, 湘潭大学计算控制技术与控制理论应用,2003:1-3.
[5] Deploying the Integrated Metropolitan ITS Infrastructure FY2000 Report,FHWA,U. S. Department of Transportation , 2001.
[6] FHWA FTA, NHTSA, FMCSA. 2002 Intelligent Transportation Systems(ITS) ProjectsBook. HTML Version. U. S. Department of Transportation ITS Joint Program Office,2002.
[7] 陈旭梅,于雷,郭继孚等,美国智能交通系统 ITS 的近期发展综述,中外公路,2003 Vo1.23, No.2: 9-120.
[8] 刘馨月,复杂背景下车牌分割技术的研究,[硕士学位论文],吉林长春,东北师范大学,2006:2-5.
[9] Barroso J,Bulas-Cruz J & Dagless E L,Real-Time Number Plate Reading[J]. 4th IFAC Workshop on Algorithms and Architectures for Real-Time Control,1997.
[10] R.Parisi. Car Plate Recognition by Neural Networks and Image Processing[J]. IEEE International Symposium on Industrial Electronics, 1998:605-610.
[11] C.H.Coetzee,et al. PC Based Number Plate Recognition Systems, In Proc.IEEE International Symposium on Industrial Electronics. 1999,1(1):605-610.
[12] C.Rittscher, J.Kato, S.Joga, and A.Blake., A Probabilistic Background Model for Tracking. 2000, 2(2): 63-67.
[13] 刘效静,成瑜.,汽车牌照自动识别技术研究,[J].南京航天航空大学学报, 1998,30(5):573-575.
[14] 牛欣,沈兰荪,基于特征的车辆牌照定位算法,[J].交通与计算机,. 2001,1(81):31-33.
[15] 刘智勇,刘迎建,车牌识别(LPR)中的图像提取及分割,[J].中文信息学报,2000,14(4):29-33.
[16] 郁梅,郁伯康,郑义,基于视觉的车辆牌照检测,[J].计算机应用研究. 1999(5):65-67.
[17] 廖金周,宣国荣,车辆牌照的自动分割,[J]. 微型电脑应用,1999(7):32-34.
[18] 戚飞虎,叶芗芸,李卫东等,机动车辆自动收费系统,[J].交通与计算机,1997,15(1):38-41.
[19] 戴青云,余英林,一种基于小波与形态学的车牌图像分割方法,[J].中国图形图像学报,2000,5A(5):411-415.
[20] 万国红,王敏,黄心汉等,基于神经网络的汽车牌照自动识别技术研究,[J]. 计算机工程与应用,2002,(6):204-205.
[21] 赵雪春,戚飞虎,基于彩色分割的车牌自动识别技术,[J].上海交通大学学报,1998,(10):3-9.
[22] 张引 , 潘云鹤,彩色汽车图像牌照定位新方法, [J]. 中国图像图形学报 .2001,6(4):374-377.
[23] 何斌,马天予,王运坚等,Visual C++数字图像处理,[M].北京:人民邮电出版社,2002.
[24] David J.Kruglinski, Scot Wingo, George Shepherd 著 , 希 望 图 书 创 作 室译.ProgrammingVisual C++6.0 技术内幕,[M].北京,北京希望电子出版社,2001.
[25] 苏明金,刘宏,刘波,Matlab 高级编程,电子工业出版社,2005:45-79.
[26] 陈永超,李汉强,Matlab7.0 与 VC6.0 的接口技术,电脑知识与技术,2006.1:129-130, SSN 1009-3044.
[27] 王琴,车辆牌照提取与识别算法的研究,[硕士学位论文],武汉,武汉理工大学,2005:11-15.
[28] 朱志刚,数字图像处理基础,[M].北京,清华大学出版社,1996.
[29] 王强,马利庄,图像二值化时图像特征的保留,计算机辅助设计与图形学学报,Vol.12,No.10,Oct,2000.
[30] N.Ostu,A threshold selection method from gay level histogram, EEE Trans. Sys.Man,Cybern,9(1),1979:62-66.
[31] 王润生,图像理解,国防科技大学出版社,1995
[32] J.Bernsen,Dvnamic thresholdingof gay-level images,Proc.of 8th Int'l Conf. onPatt. Recongn,1986:1251-1255.
[33] Byeong Rae Lee, Kyungsoo Park, Hyunchul Kang ,et al. Adaptive Local Binarization Method for Recognition of Vehicle License Plates,IWCIA 2004, LNCS 3322, Springer-Verlag Berlin Heidelberg 2004:646–655.
[34] 张铁弓,基于 DSP 的微光视频图像实时处理研究,[硕士学位论文],武汉,武汉理工大学,2002:33-36.
[35] 胡小锋,赵辉,《Visual C++/MATLAB 图像处理与识别实用案例精选》,[M]. 北京,人民邮电出版社,2004.
[36] Marr D, HILDR ETH E, Theory of edge detection[J]. Proc.Royal Soc. London, B,1980,(207):187-217.
[37] 付仲良,黄书强,陈眠等.货车图像车牌区快速定位及字符切割算法.计算机工 程与设计,2003, Vol.24 No.l:77-79.
[38] T Randen, J.H Husoy. Filtering for texture classification: a comparative study, IEEE Trans.Pattern Analysis and Machine Intelligence, 1999, Vo1.21 No.4.
[39] 陈昕,基于纹理分析的车牌图像区域分割技术的研究,[硕士学位论文]合肥,合肥工业大学 2003:17-32.
[40] 汤德佑,黄元江,基于灰度跳变的车牌自动识别技术,株洲工学院学报,Mar. 2004 Vol. 18 No. 2: 1-2.
[41] 熊军,高敦堂,沈庆宏等,基于字符纹理特征的快速定位算法,光电工程,2003, Vol. 30 No. 2: 60-63.
[42] 林纯青,徐立亚,戚飞虎,汽车图象中字符目标的提取算法,上海交通大学学报,1998, Vol.33 No. 10:1-3.
[43] 张西宁,郑南宁,汽车牌照自动识别中的目标与背景的快速分割,信息与控制,1998: 27-31.
[44] 金玲玲,廖芹,汪刘一,汽车牌照的提取方法,华南理工大学学报,2002,Vo1.30 No.7:95-97.
[45] 杨海廷,基于纹理特征的车牌识别系统的研究与实现,[硕士学位论文],成都,电子科技大学,2005:22-27.
[46] Barroso J, Rafael A. Number Plate Reading Using Computer Vision[J]. IEEE Transaction on Vehicular Technology. 2001.11.
[47] Rittscher C, Kato J. A Probabilistic Background Model for Tracking. Eccv 2000,Vol.2.
[48] 宋 建 才 , 字 符 结 构 知 识 在 车 牌 识 别 中 的 应 用 , 电 子 技 术 应 用 ,2004,Vol.30,No.5 :188-191.
[49] 郭大波,陈礼民,卢朝阳等,基于车牌底色识别的车牌定位方法,计算机工程与设计,2004,Vol.24.No.124,No5:81-84.
[50] 王建平,姜滔,基于小波分析的汽车牌照分割,合肥工业大学学报,2002 o1.25 No.6:1139-1142.
[51] Parist R. Car Plate Recognition by Neural Networks and Image Processing IEEE International Symposium on Circuits and Systems, USA, 2000, May 31-June 3.
[52] 崔江,王友仁,车牌自动识别方法中的关键技术研究,[J]. 计算机测量与控制,2003,Vol.11 No.4:260-262.
[53] 刘彦婷,徐海兰,杨磊,车牌识别软件的改进算法,北京广播学院学报(自然科学版), 2005年12月第12卷,第4期:39-43.
[54] 芮挺,沈春林,张金林.车牌识别中倾斜牌照的快速矫正算法,[J].计算机工程 2004,30(13):122-124.
[55] 王华,丁晓青,一种多字体印刷藏文字符的归一化方法,[J].计算机应用研究,2004,(6):41-43.
[56] David Llorens, Andres Marzal, et al. Car License Plates Extraction and Recognition Based on Connected Components Analysis and HMM Decoding[J]. IbPRIA 2005,LNCS 3522,Springer-Verlag Berlin Heidelberg,2005:571–578.
[57] 张云刚,张长水,利用 Hough 变换和先验知识的车牌字符分割算法,[J].计算机学报,2004.1:130-135.
[58] 陈寅鹏,丁晓青,复杂车辆图像中的车牌定位与字符分割方法,[J]. 红外与激光工程,2004,33(1):29-33.