摘要
在分析特征提取方法的基础上,提出了一种模拟人类智能的两级分类识别算法。第一级采用描绘字符整体信息的统计特征,利用SVM分类器进行识别;第二级采用描绘字符细节信息的结构特征,依据决策表区分形近字符,使车牌中形近字符的识别率得到提高。算法对提取的统计特征值进行了优化调整,使其有效地克服了字符偏移的影响,并引入可信度评判机制,使SVM分类器的性能得以提高。
The paper proposes a two-stage classification algorithm to simulate human intelligence, based on analyzing character feature extractions. In the first stage, statistical feature of describing integrated characters information is taken as the feature extraction method, and the SVM is taken as a classifier to identify the characters. In the second stage, the extraction method of structural feature is adapted, detailed information of the confusing characters is described, and then this paper distinguishes the confusing characters according to the decision tables, thus the recognition rate of the confusing characters in the license plates can be improved. Extracted statistics characteristic value is optimally adjusted so that it can overcome the impact of character offset effectively, and reliability evaluation mechanism is taken into the recognition process in order to improve the performance of the classifier.
出处
《电子技术应用》
北大核心
2011年第4期122-125,共4页
Application of Electronic Technique
基金
国家科技部科技型中小企业技术创新基金项目(国科发计字[2009]276号)
关键词
字符识别
轮廓特征
支持向量机
特征提取
形近字符
character recognition
outline feature
SVM
feature extraction
confused characters
