Abstract: By using side information in frame and between frames, this paper proposes a video watermark algorithm in wavelet domain with side information. In the algorithm, dither modulation is adopted, and the motion feature and texture feature are used as side information to determine the adaptive quantification step of every frame. Experimental results indicate that the algorithm can implement blinding examination and the invisibility is good. It is also robust to the special attacks of video watermark such as Gaussian noise, frame average, frame delete and frame displacement.

Key words: video watermark, wavelet domain, side information, dither modulation

摘要： 利用帧内和帧间的边信息，提出一种含边信息的小波域视频水印算法。该算法以抖动调制为基础，以视频帧的运动特征和纹理特征作为边信息，自适应地确定抖动调制的量化步长。实验结果表明，该算法实现了视频水印的盲检测，具有良好的不可见性，对高斯噪声、帧平均、帧删除、帧置换等攻击具有较好的鲁棒性。

关键词: 视频水印, 小波域, 边信息, 抖动调制

CLC Number: 

• TP309