A system for segmenting and extracting paper-based watermark designs

摘要

This paper presents a framework for the digitisation, extraction, and graphical representation of paper-based watermark designs embedded in paper texture: there is a growing need for this among librarians and antiquarians to aid with classification and preservation. The system is designed to handle manuscripts with foreground interference and defects; it uses a back-lighting scanning technique combined with image processing operations rather than radioactive techniques. Hence, it is faster, cheaper, safer, and easy to use. The system prototype includes a set of image processing operations which enhance, filter, and extract the watermark shape, and automatically convert it into a graphical representation. The paper focuses also on automated processes which determine the configuration of parameters in order to allow optimal content processing, in addition to the detection of watermark chainlines. With a machine readable graphical representation of the watermark, cataloguing and indexing of these heritage resources can be enhanced with the ease of digital content retrieval functionalities exploiting the advantages of digital technologies such as distribution and preservation.