基于模板匹配和神经网络的车牌字符识别方法  

A METHOD OF RECOGNIZING CHARACTERS IN VEHICLE LICENSE-PLATES USING PATTERN MATCH AND NEURAL NETWORKS

作　　者：魏武[1] 黄心汉[2] 张起森[1] 王敏[2] 王明俊[1] 

机构地区：[1]长沙交通学院道路与交通工程系,长沙410076 [2]华中科技大学控制科学与工程系,武汉430074

出　　处：《模式识别与人工智能》2001年第1期123-127,共5页Pattern Recognition and Artificial Intelligence

基　　金：湖北省重点科学技术发展计划资助项目

摘　　要：本文提出了一种基于模板匹配和神经网络的车牌识别方法。该方法集成了模板匹配识别车牌字符和神经网络识别车牌字符的各自优势。对于字符可单独分割出来的一类车牌,本文提出了一种改进的神经网络来进行字符识别;对于字符不可分割或分割困难的另一类车牌,本文提出了一种基于四灰度加权相似函数模板匹配方法来识别字符。从而克服了单一方法很难同时识别这两类车牌中的字符的不足,同时可有效地提高车牌字符识别的识别率、识别速度或识别系统的泛化能力。实验结果表明:大多数情况下,该方法车牌字符识别率超过90%,识别时间不超过1200毫秒,能更有效识别各种车牌中的字符,能更好地满足实际系统的要求。In this paper, a method of reccgnizing characters in vehicle license-plate using pattern matching and neural networks is presented. This method integrates the advantages of pattern matching and neural networks in recognizing characters in vehicle license-plates. For vehicle license-plates in which characters can be segmented, an improved neural networks is used to recognize characters; for vehicle license-plates in which characters cannot be segmented, a pattern matching algorithm based on four-grey weighted similarity function is utilized. This method can overcome the problem that it is difficult to recognize the charaters in both types of vehicle license-plated by only using pattern matching or neural networks. It can also improve the recognition rate, reduce the recognition time and increase the adaptable ability. Experimental results show that by using this method, the recognition rate of characters is higher than 90 % and the recognition time of characters for a vehicle license-plate is less than 1. 2 second, which means that this method is of more effective recognition ability and can better satisfy the real system reqwrements.

关 键 词：车牌字符识别 模板匹配 神经网络 学习算法 

分 类 号：TP391.41[自动化与计算机技术—计算机应用技术]