车牌定位及字符切分算法的研究与MATLAB仿真  

The Research of License Plate Location and Character Segmentation Algorithm and MATLAB Simulation

作　　者：徐雪丽[1] 乌旭东 白媛 XU Xue-li;WU Xu-dong;BAI YUAN(College of Mathematics and Computer Science,Yan＇an university,Yan＇an 716000,China;Radio and TV Village Service Station,Lintong District,Xi＇an 710600,China)

机构地区：[1]延安大学数学与计算机科学学院,陕西延安716000 [2]西安市临潼区广播电视村村通服务站,陕西西安710600

出　　处：《延安大学学报（自然科学版）》2018年第3期46-50,共5页Journal of Yan'an University：Natural Science Edition

基　　金：延安大学大学生创新训练项目(D2016104)

摘　　要：车牌识别主要包括4个环节,分别是:图像预处理、车牌定位、字符切分和字符识别。运用MATLAB软件,对车牌识别算法进行仿真,实现了汽车牌的定位以及字符切分。在预处理方面,对图像进行降噪,将其转化为灰度图像;在边缘检测方面,运用Roberts等算子进行边缘检测,实验比较后,最终采取较优的Roberts算子;在车牌定位方面,对边缘检测后的图像进行数学形态的闭合和腐蚀运算,去除面积较小的部分,得到车牌位置;在字符切分方面,对车牌进行二值化处理,去除车牌上的间隔符,运用比例分割和垂直投影法混合进行字符切分。License plate recognition consists of four parts： image preprocessing, license plate positioning, character segmentation and character recognition. In this paper, MATLAB software is used to simulate the license plate recog- nition algorithm and realize the positioning and character segmentation of automobile license plate. In the aspect of image preprocessing, the image was denoised, and transformed into grayscale image, and the edge detection was carried out by Roberts and other operators. After experimental comparison, the better Roberts operator was finally adopted. In terms of the license plate positioning, mathematical morphology in edge detection after the image of the closed and corrosion, and then removed the smaller part, got the location of the license plate. In the aspect of character seg- mentation, the license plate was first binarized, the interval character on the license plate was removed, and the character segmentation was can＇ied out by combining proportional segmentation and vertical projection method.

关 键 词：车牌识别 数学形态学 车牌定位 字符分割 边缘检测 

分 类 号：TP391.41[自动化与计算机技术—计算机应用技术]