首页 | 本学科首页   官方微博 | 高级检索  
     检索      

一种新的车牌数字及字母字符识别方法
引用本文:邢博,梁德群,李文举.一种新的车牌数字及字母字符识别方法[J].辽宁师范大学学报(自然科学版),2005,28(1):56-58.
作者姓名:邢博  梁德群  李文举
作者单位:1. 大连海事大学,信息工程学院,辽宁,大连,116026
2. 辽宁师范大学,计算机与信息技术学院,辽宁,大连,116029
基金项目:国家自然科学基金资助项目(60272017)
摘    要:车牌字符识别是车牌自动识别系统的关键环节.车牌数字及字母字符识别是车牌字符识别的重要组成部分.传统的模板匹配法的缺点是在车牌图像质量退化时识别率较低.提出一种新的改进的模板匹配法.该方法定义前景匹配度、背景匹配度和整体失配度3种匹配测度进行模板匹配,匹配成功则直接获得识别结果,否则,对相似字符情形应用特征区域模板匹配法,对其他情形应用空心模板匹配法.实验结果表明,该方法对车牌字符褪色、污迹、断裂、模糊、光照不均等质量退化车牌图像具有较好的识别效果.

关 键 词:模板匹配法  车牌图像  匹配度  车牌字符识别  识别率  车牌自动识别系统  污迹  字母  数字  方法
文章编号:1000-1735(2005)01-0056-03
修稿时间:2004年11月16

A New Method for Number and Letter Character Recognition of License Plate
Xing Bo,LIANG De-qun,LI Wen-ju.A New Method for Number and Letter Character Recognition of License Plate[J].Journal of Liaoning Normal University(Natural Science Edition),2005,28(1):56-58.
Authors:Xing Bo  LIANG De-qun  LI Wen-ju
Institution:XING Bo~1,LIANG De-qun~1,LI Wen-ju~2
Abstract:Character recognition is a critical step in the automatic license plate recognition system. It is the important part of the character recognition that the recognition of numbers and letters in the license plate. The low recognition rate to the poor vehicle license plate images is weakness of the traditional template match method. In this paper, a new improved pattern match method is presented Three kinds of match measure, the foreground match measure, the background match measure and the integer mismatch degree, are defined to carry on template match. If the match succeeds , the recognition result will be attained directly, otherwise the characteristic district pattern match will be applied to the similar characters and the hollow template pattern match will be applied to other conditions. Experimental results show that the proposed method in this paper has better recognizing effect when dealing with depigmentizing, dirt, split, blurry and uniform-illuminated Number and Letter characters.
Keywords:vehicle license  character recognition  pattern match  hollow template  match measure
本文献已被 CNKI 维普 万方数据 等数据库收录!
设为首页 | 免责声明 | 关于勤云 | 加入收藏

Copyright©北京勤云科技发展有限公司  京ICP备09084417号