一种基于模板匹配的汽车牌照识别方法  

A Method of Vehicle Number Plate Recognition Based on Template Matching

作　　者：路小波[1] 凌小静[1] 黄卫[1] 

机构地区：[1]东南大学,南京210096

出　　处：《交通与计算机》2004年第6期79-82,共4页Computer and Communications

基　　金：江苏省自然科学基金项目资助 ( BK2 0 0 40 77)

摘　　要：受天气、光线等因素的影响 ,易造成车牌图像灰度不均匀 ,在车牌图像上部产生阴影 ,文章提出了一种消除阴影的预处理方法。即采用模板匹配完成字符识别 ,设计字符提取算法 ,使字符提取和识别同时进行 ,避免了字符的单独分割。针对一些易混淆的字符 ,构建神经网络 ,提高字符的识别率。Vehicle number plate characters recognition is a key technology in vehicle number recognition system. Images of number plate acquired under real conditions are often anamorphic. In this paper, a method is designed to adjust the anamorphic vehicle license plates. Images are also influenced by weather and rays, which often make gray scale inhomogeneous. A preprocessing method is presented to solve this problem. Using template matching for character recognition and designing the algorithm of character extracting can avoid isolating the characters and enhance the correctness of character extracting. To the question that some characters are easy to be confused, the authors construct BP neural networks to complete effective character recognition. Testing results verify the validity of the presented methods.

关 键 词：车牌图像 字符提取 模板匹配 阴影 汽车牌照 字符识别 识别率 构建 文章 混淆 

分 类 号：U492[交通运输工程—交通运输规划与管理] TP391[交通运输工程—道路与铁道工程]