首页 | 本学科首页   官方微博 | 高级检索  
     

手写体汉字识别问题综论
引用本文:路浩如 杨源远. 手写体汉字识别问题综论[J]. 计算机应用与软件, 1994, 11(2): 1-8
作者姓名:路浩如 杨源远
作者单位:浙江大学(路浩如),浙江大学(杨源远)
摘    要:本文在讨论了人对汉字的认知心理试验研究结果之后,认为人类的汉字识别是经由整字属性及分层结构的多种途径复合而成的。这些途径中的大多数都可以在手写体汉字的机器识别中模仿运用。由此可把汉字的机器识别方法归纳为整模式法和子模式法两个大类,对它们的有利方面和困难方面进行了讨论;对影响其性能的因素进行了分析;对识别率以及它与识别字集大小和后处理等的关系提出了评估的问题。

关 键 词:手写体汉字识别  人类认知  机器识别  整模式法  子模式法  方法评估

A GENERAL DISCUSSION FOR HANDPRINTED CHINESE CHARACTER RECOGNITION
LU HAORU YANG YUANYUAN. A GENERAL DISCUSSION FOR HANDPRINTED CHINESE CHARACTER RECOGNITION[J]. Computer Applications and Software, 1994, 11(2): 1-8
Authors:LU HAORU YANG YUANYUAN
Affiliation:Zhejiang University
Abstract:Following the discussion of some experimental results in cognitive psychology, the present paper considers that the Chinese character recognition of human beings is a compound of various routines through the global character attributes and the character structure from its hierarchical components. Most of such routines can be imitated in machine recognition of handprinted Chinese characters. The advantages and difficulties of global-pattern approaches and sub-pattern approaches in machine recognition are discussed and the main factors affecting their performance are analyzed. A general evaluation for recognition rate and its relation to character set size and postprocessing is proposed.
Keywords:Handprinted Chinese character recognition   human cognition  machine recognition   global pautern approach   subpattern approach   method evaluation.
本文献已被 CNKI 维普 等数据库收录!
设为首页 | 免责声明 | 关于勤云 | 加入收藏

Copyright©北京勤云科技发展有限公司  京ICP备09084417号