一种利用颜色信息的车牌字符分割新算法  

A CHARACTER SEGMENTATION ALGORITHM OF COLOR VEHICLE LICENSE PLATE USING COLOR INFORMATION

作　　者：刘奕[1] 陈学佺[1] 蒋治华[1] 

机构地区：[1]中国科学技术大学电子工程与信息科学系,安徽合肥230027

出　　处：《计算机应用与软件》2005年第8期98-100,共3页Computer Applications and Software

摘　　要：汽车牌照自动识别系统中由于拍摄得到的字符模糊经常造成分割效果不佳。通常采用的灰度处理方法也丢失了很多颜色信息。本文提出一种基于模糊神经网络的算法,充分利用车牌的颜色信息,根据车牌底色对不同分量进行加强,直接对彩色车牌进行字符分割。实验结果表明,该算法是有效的,尤其是对车牌上汉字的分割效果较以往算法有明显提高。The image captured by Automatic Vehicle License Plate Recognition System （AVLPRS） in complicated environment may be blurred and smudged, which makes the correct rate of the recognition descend. Moreover, the regular method of transforming the color image into gray image may lost some useful information. In this paper, a new algorithm based on fuzzy neural network is presented, which fully utilizes the color information of an image. The result of the practical experiments shows that the algorithm can effectively segment the characters from the color vehicle license plate image, especially Chinese characters.

关 键 词：计算机 数字图像处理 图像采集 颜色信息 车牌字符分割算法 汽车牌照自动识别系统 

分 类 号：TP391.41[自动化与计算机技术—计算机应用技术] U492.22[自动化与计算机技术—计算机科学与技术]