基于字符连通区域性质的字符匹配技术与车牌识别技术  

作　　者：张达[1] ZHANG Da

机构地区：[1]武汉科技大学,湖北武汉430081

出　　处：《信息技术与信息化》2017年第7期55-57,共3页Information Technology and Informatization

摘　　要：本文提出了一种基于判断字符连通区域性质来进行字符匹配的技术,以实现车牌识别。首先对RGB图像通过对颜色空间的识别,得到车牌目标,再通过矩形特征匹配法定位出车牌区域。对提取出的车牌区域进行几何校正和数学形态学处理进行降噪,再将处理过后的车牌区域进行字符分割。将分割后字符的连通区域的面积大小跟重心位置与标准字符库进行比对判断,输出相应识别结果。理论分析和实验结果表明本文提出的方法是行之有效的。在实验中,我们通过Matlab进行仿真,整体识别率达到96.8%,在大多数情况下能准确识别出车牌结果。his article puts forward a technique of character matching based on judging the quality of characters-connected region to realize vehicle license plate recognition(LPR).First a license plate object is obtained by identifying the colors space of its RGB pictures;then the licensed plate region is located by matching its rectangular features.The license plate region extracted will be geometrically rectified and be processed by mathematical morphology to denoise,and then the processed license plate region will be character-segmented.When the characters are segmented,the size of characters-connected region and its center-of-gravity position will be compared and judged with the standard character repertoire and finally the relevant recognition consequence will be output.Theoretical analysis and experimental consequences show that the method put forward in this article is valid.In the experiments,we use Matlab to simulate and the recognition rate is about96.8%.So,we can accurately recognize the license plates in most cases.

关 键 词：车牌识别(LPR) 图像处理 车牌定位 字符匹配 MATLAB仿真 

分 类 号：TP391.41[自动化与计算机技术—计算机应用技术]