Spread spectrum colour video watermarking in the DCT domain

This paper presents a new robust watermarking algorithm devoted to colour video protection. This algorithm is developed out of generalising and adapting the principles we derived for still image sequences. The mark consists of a 64 bit message modulated by means of a spread spectrum technique and is embedded into the 2D-DCT hierarchy. The right room for the mark to be embedded into this hierarchy was established according to our previous study on statistical video modelling. The method thus obtained features robustness (against the general attacks, Stirmark included), transparency (fidelity and quality) and low probability of false alarm (lower than 2 × 10 ^-16 ≅ 16 ^-14). The experimental data consist of 100 colour video sequences of about 40 s each.