基于投影特征和模板匹配的车牌汉字识别算法  

Recognizing Chinese Character in License Plate Based on Projection Feature and Template Matching

作　　者：陈大海[1] 王艺霖[2] 

机构地区：[1]中国空空导弹研究院,河南洛阳471009 [2]沈阳工业大学信息学院,辽宁沈阳100178

出　　处：《河南科技大学学报（自然科学版）》2010年第6期33-36,共4页Journal of Henan University of Science And Technology:Natural Science

基　　金：国家自然科学基金项目(60675015)

摘　　要：车牌自动识别算法中,汉字识别是识别的难点,也是车牌识别技术推广应用的瓶颈。目前车牌汉字识别算法主要有特征分析算法和模板匹配算法,特征分析算法受图像质量的影响比较大,模板匹配算法在识别率方面达不到实际应用要求。本文提出了逐级识别算法:首先利用投影法进行匹配,把可识别的字符挑选出来或进行分类。然后,利用字体和区域面积的比例进行筛选,对选出的字符进行模板匹配。通过实验证实:算法不仅速度快、鲁棒性好,匹配精度也很高。For automatic license plate recognition algorithms,the recognition of Chinese characters is the most difficult issue.Currently,the available algorithms for recognition of Chinese characters in license plate include mainly the feature analysis algorithm and the template matching algorithm.The former is influenced significantly by the image quality,while the latter cannot provide a recognition rate.This paper suggested a progressive recognition algorithm,in which characters were matched firstly by a projection method to sort out or to classify the recognizable ones.Then the recognizable characters were screened in terms of their fonts and area proportions.Finally the selected characters were template matched.Experiments demonstrate that this algorithm has a high speed,a sound robustness and a high matching accuracy.

关 键 词：车牌识别 投影特征 模板匹配 

分 类 号：TP391.41[自动化与计算机技术—计算机应用技术]