基于串行分类器的车牌字符识别

License Plate Character Recognition Based on Cascading Classifier

在特征提取的基础上,对采用Bayes分类器和支持向量机的车牌字符识别方案进行比较,提出在Bayes分类器的基础上,再利用支持向量机对Bayes分类器不易区分的车牌字符进行识别的二级串行分类器融合的改进方案,在一定程度上既克服了Bayes分类器对车牌字符识别率低的问题,又解决了支持向量机识别速度慢的问题,可滿足车牌字符识别实时性和高识别率的要求。

Base on the step of feature-extracting, this paper compares two methods of Bayes classifier and support vector machines（SVM） and presents an improved method of license plate character recognition by cascading classifier which combines the method of Bayes classifier with the method of support vector machines. Experimental results show that the method not only gets over the deficiency of tow accuracy of Bayes classifier, but also solves the slow recognizing problem of support vector machines and obtains higher accuracy in real time.