摘要
从文字识别的角度出发,依据印刷体蒙古文字的特点,选择笔划轮廓结构特征、笔划穿越特征、关键特征点、蒙古文构词法等作为蒙古文字的特征,并以这些特征来设计分类器.该分类器已经被成功地用于印刷体蒙古文字识别软件中,取得了较高的识别率,从而证明了选取这些特征是合理的、有效的.
By fully analyzing the characteristics of the printed Mongolian characters, many features of the printed Mongolian characters have been selected. These features include stroke contour structure,stroke crossing,key feature points ,rules of Mongolian word-building and so on. The classifier was designed based on these features. The classifier has successfully been applied in the recognition software of the printed Mongolian characters. A high recognition rate has been achieved to prove that the features are valid and effective.
出处
《内蒙古大学学报(自然科学版)》
CAS
CSCD
北大核心
2006年第6期694-697,共4页
Journal of Inner Mongolia University:Natural Science Edition
基金
国家自然科学基金资助项目(批准号69965001和60365001)
关键词
蒙古文文字识别
特征选择
粗分类
细分类
Mongolian character recognition
feature selection
coarse classification
fine classification