多算法集成解决联机手写汉字识别粗分类问题  

MULTI-ALGORITHM COMBINATION SOLVING COARSE CLASSIFICATION PROBLEM IN ON-LINE HANDWRITTEN CHINESE CHARACTER RECOGNITION

作　　者：征荆 丁晓青[1] 吴佑寿[1] 

机构地区：[1]清华大学电子工程系,北京100084

出　　处：《模式识别与人工智能》1998年第2期239-244,共6页Pattern Recognition and Artificial Intelligence

摘　　要：本文提出一种新的联机手写汉字识别粗分类方法,它通过对两种独立的粗分类算法进行集成,并利用细分类识别器进行反馈控制,在粗分类正确率和系统识别速度方面能够达到较好的折衷.文中对该集成方法进行了一定的理论分析,并与其它集成形式进行了理论比较以及实验验证.A novel coarse classification method for on-line handwritten Chinese character recognition is presented. By combining two independent algorithms, and utilizing a feed-back introduced from fine classifier, our method is able to reach a good trade-off between correct rate and recognition speed. We make some analyses on the performance of this method from theoretical aspect, and compare it with other cocmbi-nation forms, theoretically and experimentally. The results of both analysis and experiments show that our method has the merit of high recognition rate, high speed and being easy to adjust.

关 键 词：粗分类 联机 手写汉字识别 集成 模式识别 

分 类 号：TP391.41[自动化与计算机技术—计算机应用技术]