基于HSV颜色空间和SVM的车牌提取算法  

Algorithm of License Extraction Based on HSV Clor Space and SVM

作　　者：黄社阳 刘智勇[1,2] 阮太元[2] 

机构地区：[1]五邑大学信息工程学院,江门529020 [2]江门职业技术学院,江门529020

出　　处：《计算机系统应用》2014年第8期150-153,共4页Computer Systems & Applications

摘　　要：为了克服HSV算法在车牌提取中存在与车牌颜色相近的类似车牌区域的干扰,提出一种基于HSV颜色空间和SVM相结合的车牌提取算法,该方法能够较好的对多种车牌实现精确定位与提取.首先根据字符的边界特征和HSV颜色空间分别对蓝色和黄色车牌进行粗定位,获得几个车牌候选区;然后使用训练好的SVM分类器进行字符与非字符分类;最后根据车牌特征实现定位与提取.实验表明,该方法取得了良好的效果.In order to overcome the drawbacks from the disturbance such as the color area similar to the vehicle plate, a new method of license plate extraction based on HSV and SVM is presented in this paper, which can locate and extract some kinds of plate perfectly. Firstly it caculates the coarse positioning of the blue and yellow plates based on the edge characteristics and HSV color space to get Several candidate regions. Then a pre-trained SVM is used to classify character and non character. Finally the plate was precisely located and extracted through the characteristics of the plate. Experimental results show that the proposed method has a good performance.

关 键 词：HSV颜色空间 连通区域 车牌候选区 车牌提取 SVM 字符分类 多种车牌 

分 类 号：TP391.41[自动化与计算机技术—计算机应用技术]