Differential watermarking scheme of stereo video

As stereo video become more and more popular the copyright protection for stereo video become more and more important. For stereo video, besides the normal video attack, it is easy to transfer it to a 2D video to pirate the copyright, while current stereo watermark scheme ignore this requirement which only can extract the watermark if both views are exist. A new stereo video watermarking method is presented in this paper, which can extract the watermark from either one of the two single views or the stereo video. For each view, the scheme tried to alter the 8*8 DCT coefficients of each view to embed the watermark, while for the same watermark the modifications of the two single views are in opposite directions. A new DCT coefficients relative relation between the two single views can be inferred from the DCT coefficients relation within the two single views. The experiment results show that the relation between two views is more stable than the relation of single view when the video suffered attack. We also use the Watson model to preserve the perceptual transparency of the watermarked video. The experiment results show that, the watermark scheme can satisfy both the perceptual transparency requirement and also the robustness requirement.