Résumé
The Dirty Paper Trellis Codes (DPTC) watermarking is a very efficient high capacity scheme. Nevertheless, it suffers of two drawbacks: its security weakness and its CPU computation complexity. In a previous work, we treat those drawbacks by proposing a more secure embedding space and a faster embedding technique. In the current paper, we study the robustness of the scheme with the use of a psychovisual mask and a correcting code. First, we explain how to integrate the psychovisual shaping inside of the DPTC watermarking scheme. Second, we treat of the correcting code integration. Two different masks are used, and robustness comparisons are drawn between the previous DPTC and the two protected psychovisual DPTC. One can observe that for a given SSIM degradation, the robustness results are better for the two protected psychovisual DPTC approaches.