车牌字符分割算法的比较研究  

Algorithms for Segmentation of License Plate

作　　者：刘明军[1] 谢宏霖 孙雪松[1] 由枫秋[3] 

机构地区：[1]济南大学信息科学与工程学院,山东济南250022 [2]山东英才职业技术学院,山东济南250104 [3]烟台大学计算机学院,山东烟台264005

出　　处：《济南大学学报（自然科学版）》2006年第3期245-248,共4页Journal of University of Jinan(Science and Technology)

基　　金：山东省优秀中青年科学家奖励基金(01BS01)

摘　　要：字符分割是车牌照自动识别系统中的关键步骤,通过对比3种车牌字符的分割方法,即水平投影法、模板匹配法、聚类分析法,提出了一些新的观点。经过对不同背景、光照条件下采集到的车辆图像进行实验,可以较好地解决在复杂背景条件下车牌字符分割的问题。Character segmentation is the key component of the automatic vehicle license recognition system. This paper studied three methods to segment the characters of hcense plate： level projection method, template matching method and clustering analysis, which could successfully solves the problem of the characters＇ segmentation under complex background conditions. The performance of the system was investigated on many real car images with complex background captured under various illumination conditions. We also discuss the characteristics of these methods.

关 键 词：人工智能 字符分割 水平投影 模板匹配 聚类分析 

分 类 号：TP391.4[自动化与计算机技术—计算机应用技术]