太平洋鄰里協會會論文集=Proceedings of EBTI, ECAI, SEER & PNC Joint Meeting
出版日期
1999.02.21
頁次
387 - 393
出版者
中央研究院計算中心
出版地
臺北市, 臺灣 [Taipei shih, Taiwan]
資料類型
會議論文=Proceeding Article
使用語言
英文=English
附註項
會議地點：台北中央研究院, 主協辦單位：中央研究院計算中心
關鍵詞
Kojima, Masami; Kawazoe, Yoshiyuki; Kimura, Masayuki; 藏語佛經; 字元辨識; 物件導向法; 寫本; Tibetan Buddhist Text; Character Recognition; Object-oriented Method; Manuscript
摘要
The purpose of this study is to develop a plausible method to code and compile Buddhist texts automatically from original Tibetan scripts into the Romanized form. We extract syllable from Tibetan texts and recognize automatically the Tibetan characters. The set of Tibetan characters consists of basic 30 consonants, 76 combination characters, and 4 vowels. Despite of the limited number of Tibetan characters, there are many similar characters in shape. Therefore，to separately recognize them we apply an Object Oriented Dictionary (OOD) which is created combining the categorization and character identification procedures. From our experiment，it is confirmed that it is possible to improve the rate of Tibetan character recognition dramatically by Object Oriented Method [Ref. 1,3]. We would like to express our opinion on automatic character recognition for wooden blocked Tibetan manuscripts.
