一种改进的RGB Hough车牌校正定位算法  

An improved license plate correction and localization algorithm based on RGB Hough

作　　者：王少伟[1] 尚媛园[1] 丁辉[1] 王晨[1] 胡峻铭[1] 

机构地区：[1]首都师范大学信息工程学院,北京100048

出　　处：《光学技术》2014年第1期35-39,共5页Optical Technique

基　　金：国家自然科学基金-联合资助基金项目(11178017);北京市教委科技面上项目(KM201110028017)

摘　　要：车牌的定位对车牌识别的成功率起到决定性作用。为了通过校正倾斜的车牌提高车牌在倾斜情况下字符识别的准确率,针对小型车蓝底白字的车牌提出了一种基于RGB色彩空间与Hough变换的车牌校正定位算法。对无倾斜情况的车牌可以直接定位,对于存在明显倾斜的车牌可以实现倾斜校正并定位。对车牌无倾斜情况的定位准确率为93%,倾斜车牌的定位准确率为88%。License plate localization plays a decisive role in license plate recognmon, m olu ~ of character recognition by de-skewing, a license plate correction and location algorithm based on the RGB color space and Hough transform is proposed for the license plate of the blue background with white characters. It can locate the license plate after de-skewing for the license plate is inclined. For the license plate is not inclined, it can be located directly. The license plate localization accuracy is 93% for the license plate is not inclined, and it＇s 88% for the license plate is inclined.

关 键 词：车牌定位 HOUGH变换 RGB色彩空间 倾斜校正 

分 类 号：TP391.4[自动化与计算机技术—计算机应用技术]