本科毕业设计（论文）( 2010 届 )题 目：车牌号码识别仿真分 院:电子信息分院专 业:电子信息工程班 级:06 电子本 1姓 名:学 号:指导老师:完成时间:2010 年 4 月温州大学城市学院本科毕业设计（论文）摘摘 要要车牌识别系统在交通的智能监视和管理中有着重要的应用，近几年发展非常迅速。基于图像和字符识别技术的车牌字符识别系统也是目前国内外模式识别应用研究领域的一个热点。尽管车牌的先验知识比较丰富，但是在复杂的背景下，车牌中的字符识别仍然比较困难。目前的车牌识别系统大多是针对简单场景、单一车牌。车牌字符识别系统的关键技术包括数字图像处理、车牌定位、车牌字符分割和字符识别技术。本文对已定位好的车牌进行图像位图读取、图像二值化、字符分割、提取字符特征、BP神经网络设计等模块进行了初步的研究。在字符分割方面，分析了牌照图像二值化与标准归一化以及几何校正的各种算法。借助牌照字符固定宽度、间距的固定比例关系等先验知识实现字符的分割。在特征提取方面，将字符归一化，再采用13特征法进行字符特征提取。在字符识别方面，分析比较了常用的字符识别方法。在此基础上详细分析基于BP神经网络的识别方法。实验结果证明，所采用的BP神经网络具有良好的性能满足在复杂环境下实时识别车牌的要求，具有一定的理论和实际意义。关键词：关键词：车牌字符识别；特征提取；BP 神经网络；MATLAB 温州大学城市学院本科毕业设计（论文）IABSTRACTLicense plate recognition system has important applications in the intelligent traffic monitoring and management developed rapidly in recent years. Based on image and character recognition technology license plate recognition system pattern recognition at home and abroad is also a hot field of applied research. Although the license plate of the prior knowledge rich, but in a complex background, the license plate of the character recognition is still more difficult. Most of the current license plate recognition system is a simple scenario for a single plate. The key technologies of license plate recognition system include digital image processing, license plate location, license plate character segmentation technology. This article has been positioning for a good license plate reads the bitmap image, image binarization, character segmentation, feature extraction of characters, BP neural network design module for a more detailed study.In the character segmentation area. Analyze of the license plate image binarization with the standard normalization and geometric correction algorithms. With fixed-width character license, a fixed proportion of the relationship between the pitch prior knowledge to achieve segmentation of characters.In feature extraction. The character normalization, again using 13 features of character feature extraction method.In character recognition, analyze and compared of the common character recognition. On the basis of this detailed analysis based on BP neural networks recognition. The results show that BP neural network used good performance in a complex environment to meet real-time identification license plate requirements, with some theoretical and practical significance. Keywords:License plate character recognition;Feature Extraction;BP neural network; MATLAB.温州大学城市学院本科毕业设计（论文）II目目 录录摘 要.IABSTRACT.II1 绪论.11.1 课题背景.11.2 国内外研究现状.22 图像的预处理.42.1 图像的二值化.42.2 牌照上下边框和铆钉的去除.52.3 车牌字符细化.52.4本章小结.63 车牌字符分割.74 车牌字符特征提取.94.1字符常用的特征提取方法.94.2十三点网格特征提取方法.105 车牌字符的识别.115.1 字符识别简介.115.2 基于 BP 神经网络的字符识别 .125.3BP 神经网络的设计 .135.3.1 输入层神经元个数.135.3.2 隐含层神经元数目.135.3.3 输出层神经元个数.