In this paper we propose a new imperceptible yellow printer dot watermarking scheme for documents printed on a colour printer. The scheme takes advantage of the imperfections of the human visual system to hide thousands of yellow dots over the entire page. It takes inspiration from the methods used by laser printer manufacturers for printer identification. The novelty of our approach is in providing an automatic embedding and detection method that can survive the distortions of printing and scanning. In order to achieve this a new moving window detection method is proposed. An error correction code is employed to handle the errors that could be present following detection. The scheme is evaluated through embedding and detection experiments on different types of documents; including text, architectural drawings and a cartoon. Our scheme offers an embedding capacity of 26,190 bits per page. Experiments were conducted using error correction codes with rates of 1/2, 1/3 and 1/5, given payloads of 13,095, 8,730, and 5,238 bits per A4 page respectively. We are able to successfully recover the watermark in all documents at a rate of 1/2 and 1/5, and in all document except one at 1/3. Further experiments were conducted with a smaller dot size to evaluate the impact it has on our results. With the smaller dot size we were still able to recover the watermarks from all documents when using an error correction code with rate 1/5. The capacity offered by this approach far exceeds the capacity offered by existing binary watermarking schemes, which are robust to printing and scanning. The substantially larger capacity opens up a wider range of possible applications as well as the possibility of using more robust cryptographic techniques for authentication.

Date Published: 4 May 2010
PDF: 9 pages
Proc. SPIE 7723, Optics, Photonics, and Digital Technologies for Multimedia Applications, 77230M (4 May 2010); doi: 10.1117/12.854708

Published in SPIE Proceedings Vol. 7723:
Optics, Photonics, and Digital Technologies for Multimedia Applications
Peter Schelkens; Touradj Ebrahimi; Gabriel Cristóbal; Frédéric Truchetet; Pasi Saarikko, Editor(s)