FSVM在有限集脱机手写体汉字识别中的应用  

Application of FSVM for Limited-set Off-line Handwritten Chinese Characters Recognition

作　　者：童学锋[1] 石繁槐[1] 

机构地区：[1]同济大学计算机科学与工程系,上海200092

出　　处：《计算机工程》2003年第13期109-111,共3页Computer Engineering

摘　　要：模糊支持向量机方法解决了多类支持向量机方法中的不可分区域问题。将模糊支持向量机方法引入到有限集脱机手写体汉字识别中，并以同济大学成绩自动识别系统为背景进行了一系列实验，结果表明在相同的条件下可以达到比支持向量机方法更为理想的识别效果。FSVM resolves the problem of unclassifiable regions in multi-class support vector machines. This paper applies the FSVM method to limited-set off-line handwritten Chinese characters recognition and designs a series of experiments based on the Grade OCR system of Tongji University. The result shows that one can get better recognition rate by method of FSVM than that of SVM under the same conditions.

关 键 词：支持向量机 模糊支持向量机 手写体汉字识别 FSVM 

分 类 号：TP391.43[自动化与计算机技术—计算机应用技术]