不定长车牌字符分割算法  

Method of Character Extraction in Unsteady Length Vehicle License Plate Recognition

作　　者：刘源根[1] 莫国良[2] 张三元[1] 叶修梓[1] 

机构地区：[1]浙江大学计算机学院,浙江杭州310027 [2]浙江大学城市学院,浙江杭州310015

出　　处：《现代机械》2006年第3期17-18,37,共3页Modern Machinery

基　　金：国家自然科学基金重点资助项目(60333010);国家自然科学基金资助项目(60473106)。

摘　　要：车牌号码识别包括牌照定位,字符分割,字符识别三个主要过程。字符分割是在牌照定位后把字符从牌照中分割出来的过程,然后送到字符识别系统识别。本文提出的字符分割算法在传统的垂直投影法字符分割的基础上进行了改进,对于字符个数不定、车牌字符之间有大间隔的车牌也取得了理想的字符分割效果。Plate Location, Character Extraction, Character Recognition are the three main processes of vehicle License Plate Recognition （LPR）. Character extraction is to get the characters from the plate after the plate location, and then send them to character recognition system to recognize. The proposed method in this article is based on traditional vertical projection character extraction method, but it shows better performance to plate photos with unsteady length and gaps than traditional methods.

关 键 词：车牌识别 字符分割 字符识别系统 

分 类 号：TP391.41[自动化与计算机技术—计算机应用技术] U491[自动化与计算机技术—计算机科学与技术]