车牌字符的一种精判别识别方法  

A Method of Discerning Precisly in Plate License Character Recognition

作　　者：童剑军[1] 邹明福[1] 

机构地区：[1]中国科学院自动化研究所国家文字识别工程中心,北京100080

出　　处：《计算机工程与应用》2005年第2期223-226,共4页Computer Engineering and Applications

摘　　要：该文归纳了在车牌识别系统中出现的识别错误率较高的相似字对,并针对低分辨率的特点提出了一种“子区域权值模板”车牌字符精判别方法,即将车牌字符经过二次立方插值法进行归一化处理后,分为6个子区域,针对不同的相似字赋予不同的子区域权值模板,再进行进一步的相似字精判别,以提高系统的整体识别率。试验数据分析表明,这种“子区域权值模板”精判别方法提高了识别系统的识别率。This text sums up some similar words with high discern wrong rate in Plate License Recognition(PLR)system,and puts forward a method of discerning precisly in Plate Number Character Recognition which is named as″sub -regional right value template″according to the characteristics of similar words.Having been normalized with quadratic cubic interpolation,the character's image is divided into six sub-regional.According to different similar words,we give different template with different rights in each regional and recognized the words in order to elevation the whole recognition rate.From the data of test,we can get the result that the method is effective in plate license character recognition.

关 键 词：字符识别 车辆牌照 子区域 权值 模板 

分 类 号：TP391.4[自动化与计算机技术—计算机应用技术]