基于FCM颜色聚类的车牌定位方法  

A novel approach for vehicle license plate location based on FCM color clustering

作　　者：刘德山[1] 赵颖[2] 

机构地区：[1]辽宁师范大学计算机与信息技术学院,辽宁大连116081 [2]辽宁工业大学计算中心,辽宁锦州121001

出　　处：《微型机与应用》2011年第6期45-49,共5页Microcomputer & Its Applications

基　　金：辽宁省教育厅科学研究项目(2008366)

摘　　要：提出了一种基于FCM颜色聚类的车牌定位方法。首先应用高斯差分算子对图像进行二值化;其次进行中值滤波;然后利用形态滤波,基于车牌的结构特征进行车牌的粗定位;最后基于FCM颜色聚类进行车牌的精定位。对各种条件下采集的250幅车辆图像进行实验,定位率在98%以上,同时该算法对光照影响有很好的鲁棒性。Locating the vehicle license plate plays an important role in a vehicle license plate automatic recognition system. This paper proposes a new method for vehicle license plate location based on FCM color clustering. Firstly, Difference of Gaussian （DOG） operator was used to gain binary car images; secondly ,the median filter was operated, and then rough location of the license plates can be confirmed by the morphology filter and the structure feature; finally accurate location of the license plates can be obtained by the color clustering algorithm based on Fuzzy C-Means （FCM）. The experimental results of more than 250 car images show that the proposed method is more accurate and robust to illumination variation, and the license plate location rate is higher than 98%.

关 键 词：车牌定位 高斯差分 FCM 颜色聚类 结构特征 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]