基于滴水算法的印刷体维吾尔文切分方法  

Segmentation Method of Printed Uyghur Based on Drop Fall Algorithm

作　　者：朱兰[1] 袁保社[1] 余伟[1] 

机构地区：[1]新疆大学信息科学与工程学院,新疆乌鲁木齐830046

出　　处：《计算机技术与发展》2015年第7期107-110,115,共5页Computer Technology and Development

基　　金：工信部2009年度电子信息产业发展基金项目(工信部财[2009]453)

摘　　要：为进一步提高印刷体维吾尔文字符的切分准确性,尤其针对每一行内的两个连体段之间(包括两个独写字母之间以及连体段与独写字母之间的情况)存在重叠现象时的切分,现有的方法未能有效地处理该情况,所以针对此问题提出了一种基于改进的滴水算法的切分方法。该方法首先判断两个连体段之间的关系,若存在空白间隙,则选择空白间隙的左右端作为切分点;若存在重叠现象,则选择基线空白间隙的中点作为切分点,然后根据滴落规则对两个连体段进行切分。实验结果表明,该方法能有效提高切分准确率。In order to further enhance the segmentation accuracy of printed Uyghur characters, especially for the segmentation wbere there is an overlap between two conjoined sections （ including the situation between two independent characters as well as between the con- joined section and the independent character） within each line, the existing methods have not dealt with the situation effectively. There- fore, a segmentation method based on improved drop fall algorithm for this problem is proposed. The method first determines the relation- ship between the two conjoined sections. If there is a gap, then select the left and right ends of the gap as the segmentation points. If there is an overlap,then select the midpoint of the baseline of the gap as the segmantation point,and then segment the two conjoined sections according to the dripping rules. The experimental results show that this method can effectively improve the segmentation accuracy.

关 键 词：印刷体维吾尔文 切分 连体段 重叠 基线 滴水算法 

分 类 号：TP391.43[自动化与计算机技术—计算机应用技术]