A robust video watermarking in motion vectors

This paper presents a robust video watermark algorithm in the motion vectors. Firstly. The Y component of P frame is categorized into high-texture area and low-texture area. The motion vectors are altered according to the texture of the area. Secondly, the prediction errors of the matched blocks are calculated again according to the changed motion vectors. Finally, the new motion vectors together with new prediction errors are encoded into compressed bit-streams. This algorithm can reduce the flaws and block effects of watermarked video. The experiment results show that this watermarking algorithm is robust against classical video processing such as filtering, rate changing and lossy compression.