摘要
本文针对《辞海》收入的全部一万八千余汉字,面向结构识别方法,具体给出了一种手写印刷体汉字识别字库的构建思想、原理及其生成技术.内容包括汉字形体分析.吸引分叉网表示法,字库构成和自动生成程序,实际结果表明,字库无论是空间代价还是时间代价部十分理想,完全达到了实用要求.
From the structure-oriented method, this paper proposes the principle and concrete techniques of building a dictionary library for recogniting handwritten Chinese character. The content includes the analysis of Chinese character's shapes, the representation of attracting-bifurcating net, the structure of our dictionary library, which contains all of 18000 Chinese characters in Ci Hai, and its automatic generating program. The testing results show that the library satifies practical requirment, not only in space performance but also in time performance.
出处
《杭州大学学报(自然科学版)》
CSCD
1995年第2期158-163,共6页
Journal of Hangzhou University Natural Science Edition
基金
浙江省科技计划项目(计划号92-2-091)
关键词
手写印刷体
汉字识别字库
计算机图形学
handwritten chinese character recognition
dictionary library
attracting-bifurcating net
structure analysis method