Content-dependent spatio-temporal video watermarking using 3-dimensional discrete cosine transform

In this paper we propose a content-dependent spatio-temporal watermarking scheme for digital videos. Content dependency is achieved by incorporating the hash of the video sequence into the watermark. The video sequence is treated as a 3-dimensional spatio-temporal signal for the purposes of video hash computation and watermark embedding and detection. Our experiments show that the video hash algorithm has good discriminating power and robustness against various attacks. The watermark is also shown in the experiments to have good robustness against a variety of attacks, in particular when the watermark is copied from one video sequence to another.