基于图像模糊度与主成分分析的车牌汉字识别  

License plate Chinese character recognition algorithm based on image fuzzy degree and PCA

作　　者：磨少清[1,2] 刘正光[1] 张军[1] 

机构地区：[1]天津大学电气与自动化工程学院,天津300072 [2]广西民族大学物理与电子工程学院,广西南宁530006

出　　处：《光电子．激光》2010年第3期444-447,共4页Journal of Optoelectronics·Laser

基　　金：天津市公安交通局科研基金资助项目(公科[2005]16号)

摘　　要：针对车牌汉字字符结构复杂且图像品质差异大而导致识别率不高的情况,提出了一种基于图像模糊度的主成分分析(PCA)子空间车牌汉字字符识别方法。首先通过三角模和非模糊基数计算字符图像的模糊度,然后根据模糊度将训练样本分成不同的子集并生成相应的PCA子空间族,最后以待识别字符的模糊度为依据选择相应的子空间族进行识别。实验数据表明,本文方法使得子类的类内距离变小类间距离增大,从而可以获得较高的识别率。与其他算法的对比实验进一步表明,本文算法能更好地同时满足精度和实时性的要求,具有良好的综合性能。Aiming at the low recognition rate of Chinese characters in license plates a new recognition algorithm is proposed based on image fuzzy degree and principat component analysis（PCA） subspace classification method.First,the image fuzzy degree is computed according to triangular norm and non-fuzziness cardinality.Then the training samples are divided into three subsets in terms of fuzzy degree,and their corresponding PCA subspaces are constructed.At last,the character is recognized in the specific subspaces which are chosen by the character′s fuzzy degree.The experiment results show that this algorithm can reduce the intra-subclass distances and augment the inter-subclass distances,consequently it has higher recognition rate.The comparative test results indicate further that our algorithm has more favorable comprehensive performance than other algorithms,which can better meet the demands of precision and real-time processing simultaneously.

关 键 词：车牌汉字识别 主成分分析(PCA) 模糊度 子空间族 

分 类 号：TP391.4[自动化与计算机技术—计算机应用技术]