An Improved Orthogonal Dirty-Paper Coding Method for Watermarking Applications

This paper presents an improved orthogonal dirty-paper coding method for digital watermarking applications such as multimedia data authentication or digital rights management The dirty-paper coding is based on the orthogonal Walsh-Hadamard Codes with Binning. Codeword selection is based on maximum correlation to the encoder side information. In contrast to the conventional dirty-paper codes, the encoder side information is based on a constrained extraction of the host data. The decoding error rate expression of the proposed dirty-paper watermarking scheme is derived and verified by simulations to be fairly accurate. It is found that this watermarking method is able to achieve very low visual distortion to host image and very good detection probability for the embedded watermark information.