图像背景下的满文文字提取

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。

Manchu Text Extract Under the Image Background

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.