A novel video watermark technique in motion vectors

In this paper, a novel digital video watermarking method based on motion vectors of P-frame is proposed. Rotating attack can be resisted efficiently by using this method and only half of the points need to be searched. In this method, the region information of the motion vector is used to hide copyright information. In the process of encoding, every macroblock search its best match block in the special region by the watermark. Then in the process of decoding, we get the watermark by the region of every motion vector. According to the experimental results, the PSNR is almost the same after embedding watermark. And the bit rate increase about an average of 4%. It indicates that embedding watermark in motion vector associated with region information has the advantage of little impact on the video quality, little impact on the video bit rate, capability to control the watermark capacity to be embedded, and can reduce the time of motion estimation. Based on this approach, another method of increasing watermark capacity is proposed in this paper.