Online multi-sprites based video watermarking robust to collusion and transcoding attacks for emerging applications

摘要

The rapid growth of network transmission of numeric videos has caused an urgent need for copyright protection against pirating. Different video watermarking techniques have been proposed to resolve this problem of ownership identification. Unfortunately, these techniques have been lacking efficiency to resist collusion attack. This paper proposes a novel dynamic approach for video watermarking using SURF (Speeded Up Robust Feature) based multi-sprite to protect the copyright of digital video. The proposed approach maximizes the compromise between invisibility, capacity, robustness against most important attacks and a reduced complexity. First, multi-sprite are generated to provide collusion and compression robustness. Then, the three YUV color space components of each sprite are marked using blind spatial algorithm that improves signature invisibility and enhances robustness. Experimental results illustrate good robustness against video transcoding, collusion and against additional several other attacks such as geometric manipulation (scaling, rotation, translation, cropping) and temporal attacks (frame dropping, frame swapping and frame rate changing). Besides, the suggested technique significantly reduces the processing time thanks to the proposed multi-sprites generation method.