神经网络用于多模式分类

本文叙述一种改进型HAMMING网在印刷汉字文本识别实用系统中作为粗分类的应用.给出了以3755印刷汉字为多模式分类对象的神经网络分类器的结构及其相应的算法.该方法在微型机上用软件仿真得以实现.取得令人满意的结果.

This paper presents a kind of improved HAMMING NET which is applied to practical system ofprinted Chinese text recognition as a rough classifier. The structure and relative algorithm of this neuralnetwork classifier, which deals with the multi-mode classification for 3755 Chinese Characters, are given.This algorithm is implemented by software simulation on a PC microcomputer and satistactory result isachieved.