一种多分类器集成的手写体汉字识别方法  

A Handwritten Chinese Character Recognition Method Based on Multi-Classifier Ensemble

作　　者：朱敏觉[1] 朱宁波[1] 袁异[1] 

机构地区：[1]湖南大学计算机与通信学院,湖南长沙410082

出　　处：《计算机工程与科学》2009年第4期36-39,共4页Computer Engineering & Science

摘　　要：多分类器集成是手写体汉字识别领域的新方向。本文提出的多分类器集成方法通过改进的欧氏距离分类器将待识别汉字分类到某个粗分结果集中,然后根据粗分结果集选择1-N(one-against-rest)的SVM分类器对待识别汉字进行细分,最后用贝叶斯集成两级分类器。实验对国标一级汉字中的1034个手写汉字进行识别,证明了方案的有效性。Multi-classifier ensemble is a new research direction. A new method of multi-classifier ensemble is presented in this paper. The method uses two-level classifiers to ensemble. An improved Euclid distance classifier at the first level is used to make a rough set for the test samples, and then the one-against-rest SVM classifiers are applied to recognize the handwritten Chinese more exactly. Finally the two-level classifiers are ensembled by the Bayesian law. The experiment for the handwritten national standard Chinese validates the effectiveness of the scheme.

关 键 词：手写体汉字识别 欧氏距离 支持向量机 贝叶斯集成 弹性网格 方向特征提取 

分 类 号：TP391.43[自动化与计算机技术—计算机应用技术]