摘要
支持向量机(SVM)是由Vapnik等人提出的一类新型机器学习方法。该文在字符特征提取基础上,应用SVM算法对车牌中的英文字符进行识别,克服了一般的SVM算法识别数字位图时缺乏对相邻空间像素相关性考虑的不足,在满足实时性的条件下获得高识别率。通过与基于字符特征的BP网络识别方案相比较表明,该方案性能远优于神经网络的性能,可很好地解决神经网络方法中无法避免的局部极值问题。实验讨论了在应用SVM算法对字符进行识别时,核函数K和惩罚因子C的选择对识别率的影响问题。
Support vector machine (SVM) is a kind of novel machine learning method proposed by Vapnik. A novel method of character recognition in vehicle license-plate based on SVM is proposed in the paper. This method, firstly, extracts features of characters, then uses SVM to train these features and to recognize them, getting over the overlooking of the correlation with neighbor pixels, the deficiency of traditional SVM's application in numerals recognition, obtains higher accuracy in real time applications. Comparing this method with BP network shows that this new method can avoid the problem of the local optimal solution of BP network, works better than BP network. In this experiment, the problem of the choice of kernel function and parameter C-the penalty term for misclassification is discussed.
出处
《计算机工程》
CAS
CSCD
北大核心
2003年第5期192-194,共3页
Computer Engineering