基于FCM和组合神经网络的车牌汉字字符识别方法

提出了1种组合神经网络结构的车牌汉字识别方法,主要从特征选取和分类器设计2方面研究车牌汉字字符识别,识别系统由2层神经网络组成,应用FCM算法对汉字进行粗聚类,聚类结果作为后续网络的先验知识,产生网络训练目标,采用LVQ网络进行粗分类,通过BP网络进行细分类。该种分层结构缩减了待识别模式的搜索范围,克服了传统单层识别系统识别率不高和组合网络粗分类率低的缺点。实验结果显示,本方法的识别率高,识别效率较好。

License Plate Chinese Character Recognition Method Based on Fuzzy C-means(FCM)and Combined Neural Network

One method of license plate character recognition based on combined neural network is presented in this paper.Two aspects including feature selection and classifier design are explored.Recognition system consists of a two-layer neural network.Firstly,the fuzzy C-means（FCM） algorithm is applied to coarsely clustering Chinese characters and the clustering results are used as the prior knowledge for the follow-up network.Training objectives of the neural networks are therefore produced in this way.The LVQ network is trained to classify Chinese characters at a coarse level,whereas the BP network is trained to accurately classify them.Such a hierarchical model proposed in this paper can effectively reduce the searching domain of the patterns to be recognized.In addition,the proposed model can overcome the disadvantages of monolayer and conventional combined neural network of low recognition rates and accuracy.The experimental results show that the performance of the proposed method is better than that of the traditional method in terms of recognition accuracy and efficiency.