A M.A.P. identification criterion for DCT-based watermarking

For reliably protecting the rights of multimedia data owners, digital watermarking techniques appear to be the best solution. A digital watermark is a code which is hidden into the multimedia data for carrying information regarding the copyright. In this paper the problem of the reliable identification of the watermark, in absence of the original unmarked image, is addressed. An optimum (based on a MAP, Maximum A Posteriori probability, criterion) watermark identificator is designed, for extracting watermarks inserted in the DCT domain. The watermark identificator is based on the knowledge of the probability density function of the DCT coefficients. The results support the optimality of the identificator and prove the robustness of the approach.