Interactive System for Character Segmentation of Woodblock-Printed Japanese Historical Book Images

This paper proposes an interactive system for analyzing digitized woodblock-printed Japanese historical books. This system includes stain and smear removal, background separation, character line extraction, and character extraction by region labeling with integration and separation techniques. In addition, we propose a semi-automatic and interactive graphical user interface for analyzing digital archives of woodblock-printed Japanese historical books and extracting each character image one by one. The proposed system automatically estimates and suggests feature parameters based on one or a few sample images. The user can review and adjust the parameters via the interface. The parameters approved by the user will be applied for the rest of the pages.