Memoirs of the Graduate School of Science and Engineering Shimane University. Series A

number of downloads : ?
Memoirs of the Graduate School of Science and Engineering Shimane University. Series A 35
2001-12-24 発行

前置処理系によるOCR認字率の向上法

An OCR Character Recognition Improvement Method by Preprocessor Apprpach
[Cao, Yu]
Satou, Tadamasa
File
Description
It is convenient to apply OCR to computerizing printed materials as books. Some errors in recognition, however,would be often discovered. They may cause to take not less time for corrections. It is effective to reduce such errors in recognition with covering more range of types used in Nippon-go language. To do so it is tried to develop recognizing ruby types for off-the-shelf OCR's by the preprocessor approach for ease of the development. The preprocessor provides a function by classifying ruby types from regular ones,through checking the area range of a type. This presents the preprocessor approach and its effectiveness.

 書物なと印刷物を電子化するには,OCR文字認識が簡便である.すべての文書に対して正当な認識が行われることが理想であるが,誤認字が生じるため修正には手間がかかる.これまでの経験では,ルビ付の文書の誤認字が顕著であった.そこで,ルビ文字含む文書の認字率の改善を図ることを企て,ルビOCR機能の実現を前置処理方式によって試みた.本方式は実現の容易さを狙った簡便な方式である.OCR本体はそのままで前置処理系において文字寸法による地文とルビの自動的な識別方法を実現する.本論文では本方式の実現方法と有効性について述べる.