车牌字符快速识别关键算法研究
|
摘要
随着我国经济的迅速发展,汽车数量日益增加,而车辆管理模式相对落后,甚至一些地区仍以人工管理为主,因此如何提高交通管理的效率尤为重要。随着智能交通系统(Intelligent Transportation System, ITS)的产生及发展,各种资源开始得以充分利用,交通运输系统也得以有序化、智能化的管理。作为ITS重要的技术支柱,车牌识别(License Plate Recognition, LPR)技术开始被大量研究和推广。鉴于当前该技术的发展现状,如何在复杂背景下实时精准地识别车牌成为模式识别领域的一个研究热点。
车牌识别技术主要包括车牌定位算法、车牌字符分割算法以及字符识别算法。本文在详细研究这些经典算法的基础上,结合国内车牌的特点,加以创新并提出了一些新的算法。通过大量实验验证,与以往算法相比,新算法在识别速度和精度上都有了很大的提高。
本文对车牌识别技术展开研究,主要有以下三方面的创新：
1.分析常见的车牌结构特征,学习当前经典的车牌定位算法,在此基础上提出基于多边缘特征的车牌定位算法,其中运用了颜色降维的方法,最大限度地去除了无关的冗余颜色信息。实验结果表明,该算法能够快速、精确地定位车牌区域,且抗干扰性强。
2.在定位后获取的三值车牌图像的基础上,运用Hough变换法对车牌进行倾斜校正,运用投影法对字符进行垂直分割。该方法实验效果好,鲁棒性强,能够去除铆钉等背景噪声的干扰。
3.分析、比较现有的经典车牌字符识别方法,提出了基于sift模版匹配与BP神经网络相结合的方法,分别对汉字和数字(字母)进行识别。通过仿真分析,该方法保证识别精度的同时,提高了识别速度。
With the rapid development of national economy, the quantity of automobile increases rapidly in our country. But vehicle management mode has not followed the step actually, even it is still controlled by manual management in some areas, so it is very important to improve the efficiency of traffic control. Along with the emergence and development of Intelligent Transportation System(ITS), all kinds of resources can be used more efficiently, and the intellectualized transportation system can be controlled in order. As an important part of ITS, License Plate Recognition(LPR) technology has been widely studied and applied. Given the current development status of the technology, how to recognize the plate character in greater accuracy and less time-consuming under complex background, and it has became a hotspot in pattern recognize research.
LPR technology is mainly combined by some parts such as plate location, char split and char recognition etc. In this paper, based on the characteristics of the Chinese plate, we study some classical algorithms, then propose some more effective algorithms. Many experimental results indicated that compared with those existing algorithms, new method has been greatly improved in speed and accuracy.
This paper studied LPR technology, the main innovation has three aspects as follows:
1. Analyzed structural feature of common license plate, studied classical plate location algorithms, and proposed a method of license plate location based on multiple edge features in the end, it employed the method of color dimension reduction to minimize the redundant color information. Experiment showed that the plate area can be located quickly and accurately, and it also validated strong robustness.
2. Based on3-value image obtained by plate location, we used method of Hough transform to rotate some license plates which were tilt, and used projection method to segment the string. The experiment results showed the good performance.
3. Compared and analyzed those existing classical methods of character recognition, then presented the template matching algorithm based on sift feature extraction to recognize the Chinese character, proposed the method based on BP neural network to identify the letter and digit. The experimental performance showed that those methods has higher accuracy and velocity.
车牌识别技术主要包括车牌定位算法、车牌字符分割算法以及字符识别算法。本文在详细研究这些经典算法的基础上,结合国内车牌的特点,加以创新并提出了一些新的算法。通过大量实验验证,与以往算法相比,新算法在识别速度和精度上都有了很大的提高。
本文对车牌识别技术展开研究,主要有以下三方面的创新：
1.分析常见的车牌结构特征,学习当前经典的车牌定位算法,在此基础上提出基于多边缘特征的车牌定位算法,其中运用了颜色降维的方法,最大限度地去除了无关的冗余颜色信息。实验结果表明,该算法能够快速、精确地定位车牌区域,且抗干扰性强。
2.在定位后获取的三值车牌图像的基础上,运用Hough变换法对车牌进行倾斜校正,运用投影法对字符进行垂直分割。该方法实验效果好,鲁棒性强,能够去除铆钉等背景噪声的干扰。
3.分析、比较现有的经典车牌字符识别方法,提出了基于sift模版匹配与BP神经网络相结合的方法,分别对汉字和数字(字母)进行识别。通过仿真分析,该方法保证识别精度的同时,提高了识别速度。
With the rapid development of national economy, the quantity of automobile increases rapidly in our country. But vehicle management mode has not followed the step actually, even it is still controlled by manual management in some areas, so it is very important to improve the efficiency of traffic control. Along with the emergence and development of Intelligent Transportation System(ITS), all kinds of resources can be used more efficiently, and the intellectualized transportation system can be controlled in order. As an important part of ITS, License Plate Recognition(LPR) technology has been widely studied and applied. Given the current development status of the technology, how to recognize the plate character in greater accuracy and less time-consuming under complex background, and it has became a hotspot in pattern recognize research.
LPR technology is mainly combined by some parts such as plate location, char split and char recognition etc. In this paper, based on the characteristics of the Chinese plate, we study some classical algorithms, then propose some more effective algorithms. Many experimental results indicated that compared with those existing algorithms, new method has been greatly improved in speed and accuracy.
This paper studied LPR technology, the main innovation has three aspects as follows:
1. Analyzed structural feature of common license plate, studied classical plate location algorithms, and proposed a method of license plate location based on multiple edge features in the end, it employed the method of color dimension reduction to minimize the redundant color information. Experiment showed that the plate area can be located quickly and accurately, and it also validated strong robustness.
2. Based on3-value image obtained by plate location, we used method of Hough transform to rotate some license plates which were tilt, and used projection method to segment the string. The experiment results showed the good performance.
3. Compared and analyzed those existing classical methods of character recognition, then presented the template matching algorithm based on sift feature extraction to recognize the Chinese character, proposed the method based on BP neural network to identify the letter and digit. The experimental performance showed that those methods has higher accuracy and velocity.
引文
[1]安实,王宇,王健.ITS与社会经济发展协调性评价研究[J].交通运输系统工程与信息,2011,11(4).
[2]W. Wanniarachchi, D. Sonnadara, M.K. Jayananda. License plate identification based on image processing techniques, International Conference on Industrial and Information Systems,2007, p373-378.
[3]刘亮.车牌识别技术研究,合肥工业大学硕士学位论文,2010.
[4]Devijver V N. The Nature of Statistical Learning Theory[M]. New York: Springer-Verlag,1995.
[5]贾倩,葛万成,陈康力.基于轮廓结构和统计特征的字符识别研究[J].沈阳师范大学学报,2006,24(1).
[6]C.Rittscher, J.Kato, S.Joga, and A.Blake. A Probabilistic Background Model for Tracking.2000,2(2), p63:67.
[7]Cui Y, Huang Q. Automatic license extraction from moving vehicles[J]. IEEE Transaction on Pattern Analysis and Machine Intelligence.1997,7(4), p126-130.
[8]EunRyung Lee, Pyeoung Kee King. Automatic recognition of a car license plate using color image processing, IEEE International Conference on Image Processing, 1994, p13-16.
[9]贾墨峰,陈继荣.基于字符检测的车牌定位方法[J].计算机工程,2010,36(3).
[10]刘广起,郑晓虹,张晓波.基于图像纹理特征提取的车牌定位算法[J].中国图像图形学报,2005,10(11).
[11]罗辉武,唐远炎,王翊,蓝利君.基于结构特征和灰度特征的车牌字符识别方法[J].计算机科学,2011,38(11).
[12]Chuyi Song, Haili Zhao, Shizheng Zhang. License plate recognition system based on morphology and LS-SVM. IEEE conference on Granlar Computing,2008.
[13]F. Faradji, A.H. Rezaie, M. Ziaratban. A Morphological-Based License Plate Location, IEEE International Conference on Image Processing,2007, p57-60.
[14]Xiandong Zhang, Peiyi Shen, Yuli Xiao, Bo Li, Yang Hu, Dongpo Qi, Xiao Xiao, Liang Zhang. License plate-location using AdaBoost Algorithm. IEEE conference on information and automation, p2456-2461,2010.
[15]Feng Yang. Detecting license plate based on top-hat transform and wavelet transform, International Conference on Audio, Language and Image Processing, 2008, p998-1003.
[16]中华人民共和国公共安全行业标准.中华人民共和国机动车号牌[S].GA36-2007.中国标准出版社,2007.
[17]刘伟铭,赵雪平.一种基于扫描行的汽车车牌定位算法[J].计算机工程与应用,2004,40(6).
[18]黄艳国,赵书玲,许伦辉.基于纹理特征和颜色匹配的车牌定位方法[J].微电子学与计算机,2011,28(9).
[19]卢雅琴,邬凌超.基于数学形态学的车牌定位方法[J].计算机工程,2005,31(3).
[20]陈国华.基于小波分析的车牌识别研究.中山大学硕士学位论文,2007.
[21]张玲,刘勇,何伟.自适应遗传算法在车牌定位中的应用[J].计算机应用,2008,28(1).
[22]陈扬,基于数字图像处理的车牌识别关键技术研究.武汉理工大学硕士学位论文,2009.
[23]Iran Duc Duan, Duong Anh Duc, Tran Le Hong Du. Combing Hough transform and contour algorithm for detecting vehicles'license-plates, International Symposium on Intelligent Mutimedia, Video and Speech Processing,2004, p747-750.
[24]A. Yanping Bai, B. Hongping Hu, C. Fei Li, D. Wei Shuang, E. Lixia Gong. A recognition system of China-style license plates based on mathematical morphology and neural network, International Journal of Mathematical Models and Methods in Applied Sciences,2010.
[25]赵涛,杨晓莉,王绪本,张娜.一种用于车牌定位的改进BP神经网络方法[J].计算机仿真,2007,24(2).
[26]聂洪印,周卫东,刘辉.多颜色模型和综合特征下的车牌定位新方法[J].计算机工程与应用,2010,46(12).
[27]王建,刘立,王天慧.基于四元数特定颜色对边缘检测的车牌定位[J].计算机应用,2011,31(3).
[28]Raman Maini, J.S.Sohal. Performance evaluation of prewitt edge detector for noisy images. ICGST International conference on computer science and engineering,2006,6, p39-46.
[29]杨道普,马秋禾,石磊.边缘检测Prewitt算子的改进算法[J].测绘科学,2008,33.
[30]岳鹏,彭进业.一种实用的多车牌快速定位系统[J].计算机工程与应用,2011,47(17).
[31]Danian Zheng, Yannan Zhao, Jiaxin Wang. An efficient method of license plate location[J].Pattern Recognition Letters,2005,26(7), p15-19.
[32]张浩鹏,王宗义.基于灰度方差和边缘密度的车牌定位算法[J].仪器仪表学报,2011,32(5).
[33]Milan Sonka, Vaclav Hlavac, Roger Boyle. Image processing, Analysis and Machine vision(Third Edition),2003.
[34]徐立新,梁娟,查金水,王崇科.基于颜色常量的图像特征提取及其降维方法[J].计算机工程与设计,2007,28(3).
[35]赵昭.基于颜色降维的彩色图像边缘检测研究,曲阜师范大学硕士学位论文,2010.
[36]冯慧娜,白艳萍,胡红萍.一种基于颜色和灰度跳变的车牌定位方法[J].测试技术学报,2009,23(6).
[37]Jun Fang, Shuguang Dai. A vehicle license plate location and correction method based on the characteristics of license plate. IEEE conference on intelligent control and automation, p42-46,2012.
[38]Lixia Gong, Hongping Hu, Yanping Bai. Vehicle license plate slant correction based on mathematical morphology and randon tranformation. IEEE conference on natural computation, p3457-3461,2010.
[39]Mei Wang, Guohong Wang. An optimization algorithm of vehicle license plate correction based on minimum projection distance. The 9th International conference for young computer scientists, p1701-1705,2008.
[40]Yuntao Cui, Qian Huang. Extracting characters of license plates from video sequences[J]. Machine Vision and Applications,1998, p308-320.
[41]Al-Akaidi, Marwan M, Blackledge, J.M. Fractal dimension segmentation:isolated speech recognition. IEEE seminar, p41-45,2000.
[42]Yang Wang, Qianqian Wang. Image segmentation based on multi-scale local feature. IEEE conference on image and signal processing, p1406-1409,2010.
[43]Hongyuan Wang, Chunwan Yuan, Method of license plate location based on edge detection and color information. IEEE conference on transportation, methanical and electrical engineering, p1477-1480.
[44]Richard O. Duda, Peter, E. Hart, David and G. Stork. Pattern Classification(Second Edition),2005.
[45]范立南,韩晓微,张广渊.图像处理与模式识别.第一版.北京：科技出版社,2007.
[46]Dan Zhu, Xu Chen. The recognition of handwritten digits based on BP nueral network and the implementation on android. International Conference on Intellignent system Design and Dngineering applications,2013, p1498-1501.
[47]S.N. Abdullah, M. Khalid, R. Yusof. License Plate Recogniton using Multi-cluster and Multilayer Neural Networks, Information and Communication Technologies, 2006,p1818-1823.
[48]David GLowe. Object Recognition from Local Scale-Invariant Features[A]. The Proceedings of Seventh IEEE International conference on Computer Vision[C]. Corfu, Greece:IEEE Computer Society Press,1999, p1150-1157.
[49]Lindeberg T. Scale-space theory:A basic tool for analysing strutures at differnent scale[J]. Journal Applied Statistics,1994,21(2):223-261.
[50]Brown M, Lowe D G. Invariant Features from Interest Point Groups[C]. In British Machine Vision Conference, Cardiff, Wales,2002.
[51]Hopfield J.J.Neurons with Graded Response Have Collective Computational Properties like those of Two-state Neurons[J]. Proceedings National Academy of Science,1984,81, p3088-3092.
[52]Jianlan Feng, Yuping Li, Mianzhou Chen. The research of vehicle license plate character recognition method based on artificial neural network[C]. IEEE International Asia Conterence on informatics in control, automation and robotics, p317-320,2010.
[53]Jinbang Qiu, Tianci Huang, Ikenaga takeshi. A FPGA-based dual-pixel processing pipelined hardware accelerator for feature point detection part in sift. IEEE International joint conference on NCM, p1668-1674,2009.
[2]W. Wanniarachchi, D. Sonnadara, M.K. Jayananda. License plate identification based on image processing techniques, International Conference on Industrial and Information Systems,2007, p373-378.
[3]刘亮.车牌识别技术研究,合肥工业大学硕士学位论文,2010.
[4]Devijver V N. The Nature of Statistical Learning Theory[M]. New York: Springer-Verlag,1995.
[5]贾倩,葛万成,陈康力.基于轮廓结构和统计特征的字符识别研究[J].沈阳师范大学学报,2006,24(1).
[6]C.Rittscher, J.Kato, S.Joga, and A.Blake. A Probabilistic Background Model for Tracking.2000,2(2), p63:67.
[7]Cui Y, Huang Q. Automatic license extraction from moving vehicles[J]. IEEE Transaction on Pattern Analysis and Machine Intelligence.1997,7(4), p126-130.
[8]EunRyung Lee, Pyeoung Kee King. Automatic recognition of a car license plate using color image processing, IEEE International Conference on Image Processing, 1994, p13-16.
[9]贾墨峰,陈继荣.基于字符检测的车牌定位方法[J].计算机工程,2010,36(3).
[10]刘广起,郑晓虹,张晓波.基于图像纹理特征提取的车牌定位算法[J].中国图像图形学报,2005,10(11).
[11]罗辉武,唐远炎,王翊,蓝利君.基于结构特征和灰度特征的车牌字符识别方法[J].计算机科学,2011,38(11).
[12]Chuyi Song, Haili Zhao, Shizheng Zhang. License plate recognition system based on morphology and LS-SVM. IEEE conference on Granlar Computing,2008.
[13]F. Faradji, A.H. Rezaie, M. Ziaratban. A Morphological-Based License Plate Location, IEEE International Conference on Image Processing,2007, p57-60.
[14]Xiandong Zhang, Peiyi Shen, Yuli Xiao, Bo Li, Yang Hu, Dongpo Qi, Xiao Xiao, Liang Zhang. License plate-location using AdaBoost Algorithm. IEEE conference on information and automation, p2456-2461,2010.
[15]Feng Yang. Detecting license plate based on top-hat transform and wavelet transform, International Conference on Audio, Language and Image Processing, 2008, p998-1003.
[16]中华人民共和国公共安全行业标准.中华人民共和国机动车号牌[S].GA36-2007.中国标准出版社,2007.
[17]刘伟铭,赵雪平.一种基于扫描行的汽车车牌定位算法[J].计算机工程与应用,2004,40(6).
[18]黄艳国,赵书玲,许伦辉.基于纹理特征和颜色匹配的车牌定位方法[J].微电子学与计算机,2011,28(9).
[19]卢雅琴,邬凌超.基于数学形态学的车牌定位方法[J].计算机工程,2005,31(3).
[20]陈国华.基于小波分析的车牌识别研究.中山大学硕士学位论文,2007.
[21]张玲,刘勇,何伟.自适应遗传算法在车牌定位中的应用[J].计算机应用,2008,28(1).
[22]陈扬,基于数字图像处理的车牌识别关键技术研究.武汉理工大学硕士学位论文,2009.
[23]Iran Duc Duan, Duong Anh Duc, Tran Le Hong Du. Combing Hough transform and contour algorithm for detecting vehicles'license-plates, International Symposium on Intelligent Mutimedia, Video and Speech Processing,2004, p747-750.
[24]A. Yanping Bai, B. Hongping Hu, C. Fei Li, D. Wei Shuang, E. Lixia Gong. A recognition system of China-style license plates based on mathematical morphology and neural network, International Journal of Mathematical Models and Methods in Applied Sciences,2010.
[25]赵涛,杨晓莉,王绪本,张娜.一种用于车牌定位的改进BP神经网络方法[J].计算机仿真,2007,24(2).
[26]聂洪印,周卫东,刘辉.多颜色模型和综合特征下的车牌定位新方法[J].计算机工程与应用,2010,46(12).
[27]王建,刘立,王天慧.基于四元数特定颜色对边缘检测的车牌定位[J].计算机应用,2011,31(3).
[28]Raman Maini, J.S.Sohal. Performance evaluation of prewitt edge detector for noisy images. ICGST International conference on computer science and engineering,2006,6, p39-46.
[29]杨道普,马秋禾,石磊.边缘检测Prewitt算子的改进算法[J].测绘科学,2008,33.
[30]岳鹏,彭进业.一种实用的多车牌快速定位系统[J].计算机工程与应用,2011,47(17).
[31]Danian Zheng, Yannan Zhao, Jiaxin Wang. An efficient method of license plate location[J].Pattern Recognition Letters,2005,26(7), p15-19.
[32]张浩鹏,王宗义.基于灰度方差和边缘密度的车牌定位算法[J].仪器仪表学报,2011,32(5).
[33]Milan Sonka, Vaclav Hlavac, Roger Boyle. Image processing, Analysis and Machine vision(Third Edition),2003.
[34]徐立新,梁娟,查金水,王崇科.基于颜色常量的图像特征提取及其降维方法[J].计算机工程与设计,2007,28(3).
[35]赵昭.基于颜色降维的彩色图像边缘检测研究,曲阜师范大学硕士学位论文,2010.
[36]冯慧娜,白艳萍,胡红萍.一种基于颜色和灰度跳变的车牌定位方法[J].测试技术学报,2009,23(6).
[37]Jun Fang, Shuguang Dai. A vehicle license plate location and correction method based on the characteristics of license plate. IEEE conference on intelligent control and automation, p42-46,2012.
[38]Lixia Gong, Hongping Hu, Yanping Bai. Vehicle license plate slant correction based on mathematical morphology and randon tranformation. IEEE conference on natural computation, p3457-3461,2010.
[39]Mei Wang, Guohong Wang. An optimization algorithm of vehicle license plate correction based on minimum projection distance. The 9th International conference for young computer scientists, p1701-1705,2008.
[40]Yuntao Cui, Qian Huang. Extracting characters of license plates from video sequences[J]. Machine Vision and Applications,1998, p308-320.
[41]Al-Akaidi, Marwan M, Blackledge, J.M. Fractal dimension segmentation:isolated speech recognition. IEEE seminar, p41-45,2000.
[42]Yang Wang, Qianqian Wang. Image segmentation based on multi-scale local feature. IEEE conference on image and signal processing, p1406-1409,2010.
[43]Hongyuan Wang, Chunwan Yuan, Method of license plate location based on edge detection and color information. IEEE conference on transportation, methanical and electrical engineering, p1477-1480.
[44]Richard O. Duda, Peter, E. Hart, David and G. Stork. Pattern Classification(Second Edition),2005.
[45]范立南,韩晓微,张广渊.图像处理与模式识别.第一版.北京：科技出版社,2007.
[46]Dan Zhu, Xu Chen. The recognition of handwritten digits based on BP nueral network and the implementation on android. International Conference on Intellignent system Design and Dngineering applications,2013, p1498-1501.
[47]S.N. Abdullah, M. Khalid, R. Yusof. License Plate Recogniton using Multi-cluster and Multilayer Neural Networks, Information and Communication Technologies, 2006,p1818-1823.
[48]David GLowe. Object Recognition from Local Scale-Invariant Features[A]. The Proceedings of Seventh IEEE International conference on Computer Vision[C]. Corfu, Greece:IEEE Computer Society Press,1999, p1150-1157.
[49]Lindeberg T. Scale-space theory:A basic tool for analysing strutures at differnent scale[J]. Journal Applied Statistics,1994,21(2):223-261.
[50]Brown M, Lowe D G. Invariant Features from Interest Point Groups[C]. In British Machine Vision Conference, Cardiff, Wales,2002.
[51]Hopfield J.J.Neurons with Graded Response Have Collective Computational Properties like those of Two-state Neurons[J]. Proceedings National Academy of Science,1984,81, p3088-3092.
[52]Jianlan Feng, Yuping Li, Mianzhou Chen. The research of vehicle license plate character recognition method based on artificial neural network[C]. IEEE International Asia Conterence on informatics in control, automation and robotics, p317-320,2010.
[53]Jinbang Qiu, Tianci Huang, Ikenaga takeshi. A FPGA-based dual-pixel processing pipelined hardware accelerator for feature point detection part in sift. IEEE International joint conference on NCM, p1668-1674,2009.