Video watermarking scheme based on spatial relationship of DCT coefficients

In this paper, a video watermarking scheme is proposed, which has high capacity and is robust against transcoding between MPEG2, MPEG4 and H.264. In the proposed scheme, the watermark is embedded into the DCT blocks and the frames in the same GOP are watermarked with the same watermark. During the embedding, the spatial relationship of DCT coefficient between an 8*8 block and its four 4*4 sub-blocks is exploited to select out the coefficients which are affected slightly by compression and transcoding, and then, according to the textural property of each 4*4 block, the watermark is embedded into one of the selected-out DCT coefficients by even/odd quantization. In the detection, the majority voting method is used to make the watermark more robust. Simulation results show that this scheme has high watermark capacity, and it is robust to video compression and transcoding between different compression standards.