摘要
脱机手写体汉字识别具有重要的理论意义和实践价值,目前在小字符集方面取得了比较好的效果。对大字符集来说,仍在进行研究。为了解决大字符集的手写体汉字识别问题,一般采用多层分类的方法。根据汉字的繁简和字型结构,构造了五级的二叉树SVM模型进行汉字集的粗分类,给出了模型的构造方法。在每级分类识别当中,采用不同的汉字特征和核函数,利用"one-against-rest"算法进行细分类识别。仿真实验表明,该方法能对手写体汉字分级分类识别,具有较高的识别率。
The research of off- line handwritten Chinese characters has some certain theory meaning and practical value. To the characters of big set, it has been studying. In order to solve the off- line handwritten Chinese characters recognition problems of big set, a method of multilevel classification is used generically. A five hyer binary tree SVM is constructed ro do the coarse classification according to the complexity and structure of the Chinese characters, and the method is given. In the classification and recognition of each layer, the one- against - rest method is used to do the detailed classification by different features and kernel function. The emulation test indicates that the new method can yield a higher precision to the classification of the handwritten Chinese characters.
出处
《计算机技术与发展》
2009年第9期42-45,共4页
Computer Technology and Development
基金
国家实验教学示范中心项目(411101)
关键词
脱机手写体汉字
二叉树
支持向量机
多分类
off- line handwritten Chinese characters
binary tree
SVM
multi- classification
