图像背景下的满文文字提取  

Manchu Text Extract Under the Image Background

作　　者：朱满琼 李敏[3] 许爽[3] 崔艳秋[3] 丛碧辉 

机构地区：[1]北方民族大学数学与信息科学学院,宁夏银川750021 [2]大连民族学院理学院,辽宁大连116605 [3]大连民族学院信息与通信工程学院,辽宁大连116605 [4]大连民族学院计算机科学与工程学院,辽宁大连116605

出　　处：《大连民族学院学报》2014年第1期78-81,共4页Journal of Dalian Nationalities University

基　　金：辽宁省教育厅科学技术研究项目(L2012479);国家民委科研项目(12DLZ018;12DLZ001);中央高校基本科研业务费专项资金资助项目(DC120101132)

摘　　要：提出了一种图像背景下的满文文字提取方法。用数字图像处理的方法对含有图像背景的满文图片进行单字提取,并进行去噪、细化、剪枝等预处理,找到这个单字的主轴,终止点、内部点和交叉点,在主轴上找到临界点,由临界点进行笔画生长从而达到满文文字笔画提取的目的。为了验证方法的可行性,对手写体满文图像、印刷体满文图像和彩色图像中的满文分别进行了文字提取。实验结果表明该方法对满文文字提取率较高,为进一步研究满文的识别奠定了坚实的基础。Based on the structure and characteristic of Manchu, this paper elaborates a new stroke extraction method. Firstly, the digital image processing method is used to extract the Manchu character from an image background picture, perform such preprocessing methods as de - noising, thinning, pruning and so on. Then the mid - axis of Manchu character, termina- tion point, interior points and intersection points are found. Moreover, the critical point on the mid - axis is achieved. The critical point by using the stroke growth algorithm is used to sepa- rate the stroke so as to achieve the purpose of Manchu text stroke extraction. To verify the fea- sibility of the method, this paper carries out a text extraction on handwritten Manchu, printed Manchu and Manchu in the color image. Simulation results show that the method is effective in separating the Manchu stroke and will lay a solid foundation for the further study of Manchu i- dentification.

关 键 词：满文 笔画提取 笔画生长 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]