摘要
本文提出了一种基于笔划的联机手写体汉字多级识别方法,包括拐折码笔划识别方法和笔划位置的模糊匹配方法,拐折码法具有笔划识别率高,对笔划变形的容忍度大,计算简单等特点,笔划位置匹配法不要求笔顺,对书写者的限制比较少,本方法对国标二级汉字的识别率为95%,不要求笔顺,识别速度达到实时要求。
A on-line stroke based multi-level handwriting Chinese character recognition method is presented,including turn-code stroke recognition approach and stroke position fuzzy matching mehtod.There are several ad vantages with this stroke recognition method such as high recognition rate,high degree of tolerance to stroke vari ance and simplicity.The stroke position fuzzy matching position fuzzy matching method do not need stroke strokes writing order,and there are limited restriction to writer. The recognition rate of is about 95%.The recognition speed satisties the requirement of real-time recognition.
出处
《信号处理》
CSCD
北大核心
1995年第4期265-271,共7页
Journal of Signal Processing
关键词
文字识别
模式识别
汉字识别
多级分类
Character recognition, Pattern recognition, Tree classifier