A blind spatial-temporal algorithm based on 3D wavelet for video watermarking

Based on video scene segmentation and 3D wavelet transform, a watermarking procedure for digital video, with spatial-temporal embedding and blind detecting algorithm, is presented in This work. First, a watermark image as copyright sign is preprocessed with a random location matrix. According to the characteristics of the video data, the preprocessed watermark image is adaptively spread spectrum and added into the 3D wavelet coefficients of the video shot. The proposed watermark procedure needs two keys: a random location matrix and spread spectrum watermark sequence. The first one ensures the security of watermarking procedure and the second one guarantees its robustness. The experimental results demonstrate the effectiveness and robustness of our video watermarking procedure against several video degradations and watermarking attacks.