汽车车牌模糊模式识别系统
|
摘要
基于视频图像的车牌识别系统,要求能够将运动中的汽车牌照从复杂背景中提取并识别出来。本文介绍了车牌识别系统中的关键技术(车牌提取、图像预处理、特征提取、车牌字符识别等)。
本文首先采用模糊积分理论提取车牌,然后对车牌图像进行预处理,再利用自联想存储网络进行字符恢复。在特征提取中,由模糊熵作为各特征对于分类贡献的指标进行特征提取。在字母和数字识别的第一阶段,利用支持向量机将其分成两类,第二阶段将模糊模式识别和神经网络理论结合分别对每类建立识别模型。在汉字识别中,则直接将模糊模式识别和神经网络理论结合建立识别模型。利用遗传算法优化隶属函数和网络参数,再利用最速下降法局部寻优,使系统具有很好的稳定性和精确性。
测试结果表明,字母和数字的识别率可达到96%,汉字的识别率可达到95%。
The objective of automatic vehicle number plate recognition system is to accurately extract and recognize the vehicle number plate from the complicated background. In this paper, we introduce functions of number plate extracting, number plate pre-processing, character extracting and character recognition etc.
In this paper, in the first place a fuzzy integral theory is used to extract the vehicle licence, then pre-processing is carry out, and finally character is resumed by autoassociative memory. In this paper, we introduce fuzzy entropy for character distill and using fuzzy entropy as index of character for contribution of classification. First stage of recognition of letter and digit character recognition, we carry out the classification of them by the support vector machine theory, second stage, we respectively set up recognition model of every species. We present the method of Chinese characters recognition based on fuzzy neural network. At the same time, we proposed using genetic algorithm optimize membership function and parameters of network for solving a disadvantage that the steepest descent easily get into local minimum, that made system has a good stability and accuracy.
The results of test show that the recognition rate of letter and digit character is approximate to 96% and the recognition rate of Chinese letter is approximate to 95%.
本文首先采用模糊积分理论提取车牌,然后对车牌图像进行预处理,再利用自联想存储网络进行字符恢复。在特征提取中,由模糊熵作为各特征对于分类贡献的指标进行特征提取。在字母和数字识别的第一阶段,利用支持向量机将其分成两类,第二阶段将模糊模式识别和神经网络理论结合分别对每类建立识别模型。在汉字识别中,则直接将模糊模式识别和神经网络理论结合建立识别模型。利用遗传算法优化隶属函数和网络参数,再利用最速下降法局部寻优,使系统具有很好的稳定性和精确性。
测试结果表明,字母和数字的识别率可达到96%,汉字的识别率可达到95%。
The objective of automatic vehicle number plate recognition system is to accurately extract and recognize the vehicle number plate from the complicated background. In this paper, we introduce functions of number plate extracting, number plate pre-processing, character extracting and character recognition etc.
In this paper, in the first place a fuzzy integral theory is used to extract the vehicle licence, then pre-processing is carry out, and finally character is resumed by autoassociative memory. In this paper, we introduce fuzzy entropy for character distill and using fuzzy entropy as index of character for contribution of classification. First stage of recognition of letter and digit character recognition, we carry out the classification of them by the support vector machine theory, second stage, we respectively set up recognition model of every species. We present the method of Chinese characters recognition based on fuzzy neural network. At the same time, we proposed using genetic algorithm optimize membership function and parameters of network for solving a disadvantage that the steepest descent easily get into local minimum, that made system has a good stability and accuracy.
The results of test show that the recognition rate of letter and digit character is approximate to 96% and the recognition rate of Chinese letter is approximate to 95%.
引文
[1] 杨纶标,高英仪,模糊数学,华南理工大学出版社,广州,2001,66-67。
[2] 边肇祺,张学工,模式识别,清华大学出版社,北京,2002,234-237。
[3] Kim K K, Kim K I, etal.Learing-Based Approach for License Plate Recognition. In: Proc of the IEEE Workshop on Neural Network ang Signal Processing. Sydency, Australia, 2000,616-623.
[4] J.P.Marques de Sa,Pattern Recognition Concepts,Methods and Applications,清华大学出版社,2002.11,39-41。
[5] 沈会良、李志能,基于矩和小波变换的数字、字母字符识别研究,中国图像图形学报,2000,5A(3),249-252.
[6] 郭桂蓉,模糊模式识别,国防科技大学出版社,长沙,153-156。
[7] Young Sung Soh, Byung Tae chen. Design of real time vehicle identification system. Proceeding of the IEEE[C],1994.3,2147-2152.
[8] Song Jiatao.Analysis and extraction of structure features alphabetic and digital characters on vehicle license plate[J].Journal of Image and Graphics,2002.7,945-949.
[9] Liu Wei-yi,Characteristic fot matching method of realization rapily recognition of the number plate[J].J.Optoelectronics Laser,2002, 274-276.
[10] Zhang Yin.A new approach for vehicle license plate locating for color image[J].Journal of Image and Graphics,2001.6,374-377.
[11] Li xiao-ping, Study on automobile license plate indentification systems[J].Journal of Beijing Institute of Technology,2001,21(1), 116-119.
[12] Gian Luca Foresti, Vittorio Murino and Carlo Regazzoni,“vehicle recognition and tracking from road image sequences”, IEEE transactions on vehicular thchnology, Vol.48, No.1, JAN 1999.
[13] Takeo Kato utal,“preceding vehicle recognition based on learning from sample image”,IEEE transactions on intelligent transportation systems, Vol.3, No.4, DEC 2002.
[14] Yong-sheng Yang utal,“A new fuzzy approach for pattern recognition with application to EMG classification”, in IEEE 1996 Vol 5, 1109-1114.
[15] Hisao Ishibuchi, Ken Nozaki, and Hideo Tanaka,“Selecting fuzzy if-then rules for classification problems using genetic algorithms”, in IEEE transaction on fuzzy systems, Vol 3, No.3, AUG 1995.
[16] Xie Dailiang, JI Haifeng, Huang Zhiyao, Li Haiqing,“An Online Flow Pattern Identification System for Gas-oil Two-phase Folw Using Electrical Capacitance Tomography”,in IEEE 2004.
[17] Martin T.Hagan,Howard B.Demuth,Mark H.Beale,Neural Network Design,机械工业出版社,北京,2002。
[18] Matlab 6.5 辅助图像处理,电子工业出版社,北京,2003。
[19] 胡宝清,模糊理论基础,武汉大学出版社,武汉,2004。
[20] 王立新,模糊系统与模糊控制教程,清华大学出版社,北京,2003。
[21] 张志涌,精通 Matlab 6.5,北京航空航天大学出版社,北京,2004。
[22] 宋加淘,刘济林,车辆牌照上英文和数字字符的结构特征分析及提取,中国图像图形学报,2002,945-946。
[23] 刘广起,严殊,基于模糊 ARTMAP 神经网络的车牌识别,郑州工业高等专科学校学报,2004。
[24] 田盛丰,黄厚宽,李洪波,基于支持向量机的手写体相似字识别,中文信息学报,第14 卷第 3 期,38-39。
[25] 冯国进,顾国华,车牌自动定位与模糊识别算法,光电子.激光,2003。
[26] 陈海燕,戚飞虎,模糊集在复杂背景下目标选取中的应用,红外与毫米波学报,2000,381-383。
[27] 邹晓涛,陈学佺，汽车牌照自动识别系统的设计与研制，红外与激光工程，2002,416-418。
[28] 李小平,任江兴,杨德刚,车牌识别系统中若干问题的探讨,北京理工大学学报,2001。
[29] 徐全生,于霞,梁乐彬,汽车牌照图像的预处理研究,沈阳工业大学学报,2002,121-122。
[30] 李弼程、彭天强、彭波,智能图像处理技术,电子工业出版社,北京,2004。
[2] 边肇祺,张学工,模式识别,清华大学出版社,北京,2002,234-237。
[3] Kim K K, Kim K I, etal.Learing-Based Approach for License Plate Recognition. In: Proc of the IEEE Workshop on Neural Network ang Signal Processing. Sydency, Australia, 2000,616-623.
[4] J.P.Marques de Sa,Pattern Recognition Concepts,Methods and Applications,清华大学出版社,2002.11,39-41。
[5] 沈会良、李志能,基于矩和小波变换的数字、字母字符识别研究,中国图像图形学报,2000,5A(3),249-252.
[6] 郭桂蓉,模糊模式识别,国防科技大学出版社,长沙,153-156。
[7] Young Sung Soh, Byung Tae chen. Design of real time vehicle identification system. Proceeding of the IEEE[C],1994.3,2147-2152.
[8] Song Jiatao.Analysis and extraction of structure features alphabetic and digital characters on vehicle license plate[J].Journal of Image and Graphics,2002.7,945-949.
[9] Liu Wei-yi,Characteristic fot matching method of realization rapily recognition of the number plate[J].J.Optoelectronics Laser,2002, 274-276.
[10] Zhang Yin.A new approach for vehicle license plate locating for color image[J].Journal of Image and Graphics,2001.6,374-377.
[11] Li xiao-ping, Study on automobile license plate indentification systems[J].Journal of Beijing Institute of Technology,2001,21(1), 116-119.
[12] Gian Luca Foresti, Vittorio Murino and Carlo Regazzoni,“vehicle recognition and tracking from road image sequences”, IEEE transactions on vehicular thchnology, Vol.48, No.1, JAN 1999.
[13] Takeo Kato utal,“preceding vehicle recognition based on learning from sample image”,IEEE transactions on intelligent transportation systems, Vol.3, No.4, DEC 2002.
[14] Yong-sheng Yang utal,“A new fuzzy approach for pattern recognition with application to EMG classification”, in IEEE 1996 Vol 5, 1109-1114.
[15] Hisao Ishibuchi, Ken Nozaki, and Hideo Tanaka,“Selecting fuzzy if-then rules for classification problems using genetic algorithms”, in IEEE transaction on fuzzy systems, Vol 3, No.3, AUG 1995.
[16] Xie Dailiang, JI Haifeng, Huang Zhiyao, Li Haiqing,“An Online Flow Pattern Identification System for Gas-oil Two-phase Folw Using Electrical Capacitance Tomography”,in IEEE 2004.
[17] Martin T.Hagan,Howard B.Demuth,Mark H.Beale,Neural Network Design,机械工业出版社,北京,2002。
[18] Matlab 6.5 辅助图像处理,电子工业出版社,北京,2003。
[19] 胡宝清,模糊理论基础,武汉大学出版社,武汉,2004。
[20] 王立新,模糊系统与模糊控制教程,清华大学出版社,北京,2003。
[21] 张志涌,精通 Matlab 6.5,北京航空航天大学出版社,北京,2004。
[22] 宋加淘,刘济林,车辆牌照上英文和数字字符的结构特征分析及提取,中国图像图形学报,2002,945-946。
[23] 刘广起,严殊,基于模糊 ARTMAP 神经网络的车牌识别,郑州工业高等专科学校学报,2004。
[24] 田盛丰,黄厚宽,李洪波,基于支持向量机的手写体相似字识别,中文信息学报,第14 卷第 3 期,38-39。
[25] 冯国进,顾国华,车牌自动定位与模糊识别算法,光电子.激光,2003。
[26] 陈海燕,戚飞虎,模糊集在复杂背景下目标选取中的应用,红外与毫米波学报,2000,381-383。
[27] 邹晓涛,陈学佺，汽车牌照自动识别系统的设计与研制，红外与激光工程，2002,416-418。
[28] 李小平,任江兴,杨德刚,车牌识别系统中若干问题的探讨,北京理工大学学报,2001。
[29] 徐全生,于霞,梁乐彬,汽车牌照图像的预处理研究,沈阳工业大学学报,2002,121-122。
[30] 李弼程、彭天强、彭波,智能图像处理技术,电子工业出版社,北京,2004。