脱机手写体汉字识别的支持向量机方法研究  

Research on Method of Off-line Handwritten Chinese Characters Recognizing Based on SVM

作　　者：王建平[1] 张丽萍[1] 

机构地区：[1]合肥工业大学电气与自动化工程学院,合肥230009

出　　处：《计算机与数字工程》2008年第4期146-150,共5页Computer & Digital Engineering

摘　　要：提出一种采用二叉树粗分类与SVM细分类结合的分类识别方法,充分发挥SVM在二类分类问题方面相对于单一SVM方法的优势。实验结果表明提出的识别方法在解决复杂多分类脱机手写体汉字分类识别问题上有效的提高了分类精度和速度。The problems associated with complex pattern and multi - classification in off - line written Chinese characters recognition are addressed. A method of classification recognition combined with binary tree and SVM （ support vector machine） has been presented. And the generalization ability has improved greatly. The experimental results indicate that the new method yields higher precision and speeds up support vector machine multi - class classification.

关 键 词：脱机手写体汉字 多分类 二叉树 SVM 汉字分类识别 

分 类 号：TP391.43[自动化与计算机技术—计算机应用技术]