车牌识别系统中字符分割的算法

Character Segmentation Algorithm in License Plate Recognition System

车牌字符分割算法是车牌识别系统的重要组成部分.针对中国车牌的7个字符间的普遍性和特殊性,考虑倾斜和无倾斜这两种车牌图像,采用投影法并辅之以连通域法,通过图像预处理、车牌定位、字符分割三个主体流程,综合运用图像灰度化处理、边缘检测、灰度图腐蚀、平滑处理、车牌定位、车牌校正、切除边框、图像去噪处理和图像膨胀等关键技术,最终成功分割出车牌字符.实验结果表明,该字符分割算法的实时性和准确性较高,能很好地适用于倾斜和无倾斜的车牌图像,受图像拍摄环境条件的影响较小,为车牌识别系统的优化提供了一种新方法.

The license plate character segmentation algorithm is an important part of the license plate recognition system.Aiming at the universality and particularity between seven characters of Chinese license plate,tilting and non-tilting license plate images are considered,projection method as well as connected domain method is adopted.With the three main processes（image preprocessing,license plate location,character segmentation）,gray image processing,edge detection,image erosion,smooth processing,license plate location,license plate correction and frames cutting,image noise removal,image dilation and the comprehensive use of other key technologies,finally,the license plate characters are segmented successfully.Experimental results show that the character segmentation algorithm has high real？time and accuracy,and can be well applied to tilting and non-tilting license plate images,which is less affected by the environment conditions of image shooting.It provides a brand-new method for the optimization of the license plate recognition system.