一种基于SVM的车牌汉字的有效识别方法

支持向量机(SVM)是20世纪90年代初由Vapnik等人提出的一类新型机器学习方法,此方法能够在训练样本很少的情况下达到很好的分类推广能力。文章应用SVM算法对车牌中的汉字字符进行识别,在无字符特征提取的情况下可得到较高的识别率和识别速度。通过与无字符特征提取的BP网络识别系统比较表明,在小样本的情况下,该方法的识别率远优于神经网络,并避免了神经网络的局部极值等的问题。

An Effective Method for Chinese Character Recognition in License Plate

Support vector machines(SVM)is a kind of new machine learning method proposed by Vapnik and his team in1990s.This method gains a better generalization ability with a few of training examples.An application of SVM algo-rithm in License Plate Character Recognition is proposed in this paper.Without extracting features of the characters,it can obtain quick result and high recognition rate.Comparing with BP network shows this method works better than BP network at recognition rate and speed,avoids the problem of the local optimal solution of BP network and needs no empirical knowledge.