摘要
提出了一种车牌汉字识别方法.该方法基于统计特征中的投影特征将车牌汉字根据结构特征进行粗分类,对于粗分类结果建立不同的BP神经网络分类器,训练完毕后,以MATLAB为软件平台,利用网络参数对车牌汉字进行分类识别.结果表明,该方法对车牌汉字识别有效,识别率高.
This paper proposes a method for recognizing Chinese characters of license plates.Initially,the Chinese characters are roughly sorted according to structure characteristic based on projection characteristic of the statistical nature.Several different BP neural networks classifiers should be created based on the rough classification result.Then the Chinese characters can be classified with the trained network parameters ultimately using MATLAB as software platform,the results show that the method has the advantages of high accuracy.
出处
《东北师大学报(自然科学版)》
CAS
CSCD
北大核心
2012年第3期60-64,共5页
Journal of Northeast Normal University(Natural Science Edition)
基金
国家自然科学基金资助项目(50478007)
吉林省科技发展计划项目(20100458)
关键词
投影特征
结构特征
BP神经网络分类器
MATLAB
projection characteristic; structure characteristic; BP neural network classifier; MATLAB