简牍图像文字切分算法研究  

Study on Segmentation of Chinese Characters in Bamboo Scripts Image

作　　者：刘瑛[1] 王绪本[2] 

机构地区：[1]成都理工大学信息工程学院,成都610059 [2]成都理工大学研究生院,成都610059

出　　处：《科学技术与工程》2007年第21期5584-5589,5608,共7页Science Technology and Engineering

基　　金：国家十五科技攻关项目(2004BA810B)子专题(04SK5008)资助

摘　　要：简牍图像具有干扰噪声大、文字大小不一的特点,传统的文字切分算法很难准确区分图像中的文字与噪声。针对简牍图像特点,利用投影法进行文字初步切分,运用多种补偿算法对初切分结果进行去噪处理和精确定位。实验结果表明,该算法具有文字切分效果好、运算时间短、易于实现的特点,对简牍图像处理的工程应用具有积极意义。Bamboo scripts image has characteristics of big noise interference and the text on it is inconsistent,so traditional character segmentation algorithm is very difficult to accurately distinguish between words and the noise.In response to the characteristics of bamboo scripts image,preliminary segmentation of character making use of projection algorithm is first carried out,and then compensatory methods to get rid of noise and to get accurate position of the word is applied.Experimental results show that this kind of character segmentation algorithm has such feathers as it is effective in character segmentation area of bamboo scripts image,its computational time is relatively short,and it is easy to implement in software,so this method is proposed for bamboo scripts image processing with a positive sense of application.

关 键 词：简牍 切分 投影 噪声 

分 类 号：TP391.43[自动化与计算机技术—计算机应用技术]