识别在线手写汉字的模糊属性自动机

本文提出一种识别在线手写汉字笔划的模糊属性自动机,为汉字识别打下了基础.汉字的笔划由笔段构成,利用笔段的长度信息,借助模糊信息处理方法,以不变嵌入原理为着眼点,提出一种模糊属性文法及其相应的模糊属性自动机.这种文法在句法形式上是有限状态文法,但语义规则中包含上下文的信息,其能力大大超过有限状态文法,相应的自动机能有效地识别笔划.经对大量在线手写汉字的识别,表明了模糊属性自动机的实用性.

A Fuzzy Attributed Automaton for On-Line Chinese Character Recognition

A fuzzy attributed automaton for on-line recognition of Chinese character strokes is proposed. We know that a Chinese character consists of several strokes, and that a stroke iscomposed of a few linesegments. By using the length information of line segment and fuzzyinformation processing technique, a fuzzy attributed grammar as well as the correspondingautomaton are defined. The production form of the grammar is a finite-state form, butthe context information of the attributes are introduced into the semantic part of the grammar.On the basis of such fuzzy attributed automaton, a large amount of Chinese characters canbe recognized effectively.