基于支持向量机的车牌字符识别

Vehicle license plate recognition based on SVM

针对车牌图像中车牌字符的识别问题,采用支持向量机(Support Vector Machine,SVM)进行识别。首先对车牌字符图像进行预处理,生成二值化图像;然后通过各类字符样本的正样本和负样本对SVM分类器进行训练,得到每类字符的支持向量;最后利用训练结果进行未知车牌字符的识别。基于OpenCV库实现了程序原型,测试结果表明车牌字符的样本回判率为94%,未知样本的识别率为91.8%,取得了良好的识别效果。

To solve the problem of character recognition in a license plate image, the paper proposed a method using Support Vector Machine （SVM）. First, characters of license plate image were preprocessed, generating two value images; then through various charac- ter sample, including positive and negative samples, trained the SVM classifier, getting each character correspondent support vector; finally used the training results to recognized unknown plate license characters. A prototype program based on OpenCV is applied. The results show that, by using support vector machine in plate license character recognition, the sample recognition rate is 94%, the unknown sample recognition rate is 91.8%, which indicates high recognition accuracy.