Research On Digital Video Watermarking Algorithm For Resistance To Synchronization Attack

Digital video watermarking is a powerful tool for video copyright protection and security authentication. Video data is redundancy, so video watermark usually encounters synchronization attacks, such as frame dropping, cut-paste etc, how to improve synchronization attack-resistance of video watermarking has become an important research trend.To improve the robustness of video watermarking to synchronization attacks, a watermarking algorithm based on Discrete Cosine Transform(DCT) and region jointly classification is proposed in this paper. The algorithm introduces the concept that the watermark should close depend on the video content, proposes that the content features contain both watermark and its position, so the same watermark is embedded in the similar area. Besides, considering that the algorithm is sensitive to starting position, starting position scanning method is proposed to guarantee the consistence of region division between watermark embedding and retrieving.Wavelet transform provides a new approach for improving synchronization attack-resistance of video watermarking, based on Time-domain Discrete Wavelet Transform(TDWT),a localized watermarking scheme is introduced, according to robust content features. Important embedded regions are selected in both low and high frequency component to improve the spatial synchronization attack-resistance.Parity checking is used in the two algorithms to realize the watermark embedding and blind retrieving, and the performance of the two algorithms is evaluated. The experimental results demonstrate that the two proposed video watermarking algorithms have good fidelity and preferably applicability. They are robust to video synchronization attacks such as frame dropping, cut-paste and other common attacks such as noise and compression. They are feasible algorithms.