摘要
文章建构了一个基于灰度的繁体手写汉字字库建库系统,该系统是由二值化、汉字切分、二值化修正、人工修正切分结果、数据压缩存储等几部分组成。其中二值化算法在最大限度地保证字库的分辨率及质量方面起着重要作用。文章在最大类间方差法(即大津法)及局部阀值法的基础上,结合汉字切分过程的特点,提出了一种二值化修正方法。实验结果表明该方法的效果比起传统的方法有显著提高,能有效地处理字库样本中由于笔迹灰度不均匀所带来的问题。
This paper describes a traditional Chinese handwritten character sample database building system,which contains the process of thresholding,character segmentation,thresholding re-correction,manual adjustment and data compression.Thresholding algorithm plays a key role for the quality of the database.Based on an improved Otsu algorithm and local threshold method,a new thresholding re-correction method is proposed.Experiments show that the proposed method works much better and effective than conventional method, and it can solve the problem of bad sampling images caused by different pen intensity very well.
出处
《计算机工程与应用》
CSCD
北大核心
2005年第35期177-179,232,共4页
Computer Engineering and Applications
基金
国家自然科学基金(编号:60275005)
广东省科技计划(编号:2003C50101
04105938)
关键词
繁体手写汉字
二值化
二值化修正
切分
traditional Chinese handwritten character,threshold,thresholding re-correction,character segmentation
