Dynamic wavelet feature-based watermarking for copyright tracking in digital movie distribution systems

Abstract

We propose novel watermark embedding and detection methods to address the problem of copyright tracking of movies distributed to theaters by satellites. Temporal and spatial wavelet transformation as well as feature-based watermark embedding procedures are deployed such that the proposed watermarking algorithm is able to affect a compromise tradeoff between the visual quality and the robustness. The multi-resolution nature of the wavelet transformations of a video sequence makes the watermark very robust and secure. Wavelet features make huge watermark patterns possible, which allows unique labeling of huge number of videos and video sequences dynamically. Changing some of the features in the video wavelet transformation domain according to the defined rules makes the watermark detection semi-oblivious, which allows watermark detection very robust and independent without the original sequence. Our proposed watermark can also be detected robustly after many kinds of malicious pirating attacks and modifying, and experimental results are reported in this paper.