Character Image Database of Woodblock-Printed Japanese Historical Book Images

This paper describes a database designed for assisting the work of transcription of digitized woodblock-printed Japanese historical books. The database system manages information about character segmentation results and transcription data, which are obtained by using our character segmentation and transcription systems. Each record of the database contains individual character images and transcription as well as metadata, e.g. The book title, page, line and character numbers, and size of characters. The image matching function of the database helps users to search appropriate candidates for unclear character images to transcribe, which are found during the transcription process.