Abstract
ABSTRACT This paper proposes an effective approach for video watermarking. Video watermarking plays an important role in many applications such as copy control, owner identification, broadcast monitoring, and so on. The main difficulty comes from the temporal dimension because it provides useful information for attackers to remove or desynchronize the embedding watermark. We propose a video watermarking method motivated from the spatially localized image-dependent (SLIDE) sub-frame watermarking system to resist the linear collusion attacks and de-synchronization attacks. In order to enhance the robustness to other common attacks, such as averaging and compression attacks, we generate watermarks by the block-based mechanism. In addition, in order to improve the imperceptibility under the same embedding strength, we embed watermarks in multi-channels by Human Visual System (HVS). Finally, we employ a blind threshold technique to determine the detection result. Our experimental results demonstrate that the above mechanisms are effective to resist spatial average filtering attacks, temporal frame averaging attacks, and MPEG compression attacks.