Japanese / English


論文の言語 英語
著者 Masakazu Iwamura, Tomohiko Tsuji, Koichi Kise
論文名 Memory-Based Recognition of Camera-Captured Characters
論文誌名 Proceedings of the 9th IAPR International Workshop on Document Analysis Systems (DAS2010)
ページ pp.89-96
年月 2010年6月
要約 This paper addresses how to quickly recognize a characterpattern using a lot of case examples without learning. Herewithout learning means just finding the most similar examplefrom the case examples, and pretend as if the OCR understandsthe definition of the character. This strategy isexpected to work well in most cases with a large dataset,however, also expected to take a lot of time for finding themost similar example. In this paper, we show that a lotof case examples can be processed in a short time. As atestbed, we handle recognition problem of camera-capturedprinted characters. Using a database storing 100 fonts, theproposed method achieved 97.0% of recognition rate for imagescaptured from the right angle and 95.8% for those from45 deg. with 4.56ms of processing time, that is about 220characters per second including everything.