基于改进BP网络的车牌字符识别方法研究  

Research of vehicle license plate recognition based on BP neural network

作　　者：李会民[1,2] 张仁津[1] 

机构地区：[1]贵州师范大学数学与计算机科学学院,贵州贵阳550001 [2]濮阳县职业高级中学,河南濮阳457000

出　　处：《计算机工程与设计》2010年第3期619-621,共3页Computer Engineering and Design

基　　金：贵州师范大学研究生创新基金项目(研(2009)11)

摘　　要：介绍了车牌自动识别的过程和原理、车牌识别中所需要的关键图像处理技术,并对BP神经网络算法进行了阐述。结合目前车牌编制的特点,基于BP神经网络原理对车牌自动识别技术中的字符识别技术进行了新探索,提出了适用于新《机动车登记规定》车牌编制方法的神经网络结构和识别算法。通过恰当设置BP网络隐层节点数,详细介绍了改进BP网络算法步骤,分析了易混字符,并用MATLAB进行了实验,实验结果表明,改进网络结构后的方法识别准确率高、识别过程速度快、系统鲁棒性强。The process of the vehicle license plate automatic recognition,and the key technology of graphic procdure are introduced, the BP neural network algorithms is discribed.Combinating the new character of vehicle license plate,some researches based on the BP neural network are done,the new structure of neural networks meet the need of the new ruler-of-vehicle-register and introduce the new recognition algorithms.Via properly setting parameter of hiden layer,the process of the new algorithms on BP network is introduced in detail,some vague characters are analyzed.The recognition using Matlab software is simulated,and some results of simulation are provided.The experiments show that this system has a high rate of accuracy,quick speed of recognition,good robustness,and better practical value.

关 键 词：BP网络 车牌定位 字符分割 字符识别 车牌识别 

分 类 号：TP18[自动化与计算机技术—控制理论与控制工程]