Abstract: This paper presents an approach to recognize the Tibetan by analyzing the structure and geometric features of the Tibetan writing. It lets the system “learn” the geometric features of Tibetan consonant and vowel symbols, splits the word on the basis of structures of print Tibetan. The results of the segmentation are used to do the next identification. It converts them to Latin which makes people can “read” Tibetan. The experimental result shows the accuracy of this approach can be over 89%.

Key words: character recognition, Tibetan recognition, artificial intelligence

摘要： 在分析藏文的几何结构的基础上，提出了一种基于几何形状分析的藏文识别方法。通过对藏文的辅音和元音符号的几何特征提取，达到学习的目的，根据印刷体藏文的组成结构特征，对其进行切分，进而分别对切分结果进行识别，将藏文字符转译为拉丁文，使得即使不认识藏文的人也能“读”藏文。实验结果表明，该方法的正确率达89%以上。

关键词: 文字识别, 藏文识别, 人工智能