联机手写汉字识别

联机手写汉字识别

施小平

目录

目录 2

中文 摘要 关键字................................................... 3

英文 摘要 关键字................................................... 4

绪论 5

Ⅰ.笔迹采集 7

Ⅱ.预处理 8

1.字符分割 8

2.去噪和空间采样 9

3.平滑处理 9

Ⅲ.笔划提取 10

1.笔段，笔划（笔画）,自然笔画 10

2.笔段的分类和笔划的分类 11

3.笔划的提取 12

Ⅳ. 位置关系判定 18

1.位置关系判定 18

Ⅴ.单字识别 24

1.单字 24

2.单字的主要类型结构 24

3.距离准则 24

4.单字识别 26

5. 当前单字识别方法的不足和改进思路 31

Ⅵ.词组联想 32

总结 34

参考文献 36

摘 要：本文首先介绍联机手写汉字识别的研究背景、研究价值和发展现状，接着介绍了手写汉字识别的特点和相关技术，然后具体阐述了一种基于笔划提取和笔划位置关系判定的识别算法，最后用采集到的手写汉字样本，在使用JAVA编程语言实现的联机手写汉字识别系统上进行训练和测试。这种识别算法主要包括六个模块：笔迹采集、预处理（平滑、去噪等）、笔划提取、笔划位置关系判定、单字识别、词组联想。字符集合为常用的国标基本集GB2312一级、GB2312二级共6763个汉字。

关键字：联机手写汉字识别；JAVA；笔划提取；汉字判定；词组联想

Abstract: First of all, the paper introduces the research background, research value and the development state of the on-line handwritten Chinese character recognition. Next, the feature and the interrelated technology of the handwritten Chinese character recognition are discussed. Then, the paper expounds a recognition algorithm, which is based on strokes extraction and etymon judgement. And I design an on-line handwritten Chinese character recognition system coded by JAVA programming language, where some samples of handwritten Chinese characters are trained and recognized. The above algorithm contains six major modules: handwriting collection, preprocessing (smoothing, noise reduction et al.), stroke extraction, position relation judgement, individual character judgement and phrases association. The character is the first and second part of the GB2312-80 coding set, which contains 6763 Chinese characters.

Key words: on-line handwritten Chinese character recognition; JAVA; strokes extraction; etymon judgement; phrases association

绪论...