Motion Picture Watermarking Via Quantization of Pseudo-Random Linear Statistics

In this paper, we propose a semi-blind video watermarking scheme, where we use pseudo-random robust semiglobal features of video in the spatially wavelet transform domain. We design the watermark sequence via solving an optimization problem, such that the features of the mark-embedded video are the quantized versions of the features of the original video. We use a novel approach in generating the watermark such that the temporal correlation within the video frames is reflected on the generated watermark. This is achieved by i) using a special weight distribution along the time axis ii) modifying the regularization step to reflect the temporal characteristic of the host. We experimentally show the robustness of our algorithm against temporal filtering attacks and at the same time show a common weakness in 3D transform based video mark embedding methods.