Switching between orthogonal watermarks for enhanced security against collusion in video

Recent efforts for watermarking digital video basically extend the results previously obtained for still images. Thus, most of the proposed algorithms rely on a frameby-frame approach. Such a direct adaptation leads to non-secure algorithms because of collusion attacks. Depending of the targeted application, this may be a critical concern. A new watermarking system, extending current schemes based on spread spectrum, is consequently introduced in this paper. It basically embeds alternative orthogonal watermarks in successive video frames, which succeeds in obtaining superior performances in terms of security. Finally, a discussion is conducted to check that this new architecture cannot be easily broken with a straightforward attack.