针对数字视频版权保护及信息隐藏技术的需要，本文提出了一种基于帧间小波变换的视频水印算法。该算法首先对视频序列进行等长分组，然后对各组中的视频序列进行最大级数的帧间一维小波变换，从而可以得到一帧低频图像和多帧高频图像，其中低频图像聚集了视频序列的大部分能量，而高频图像的能量相对较小。本文对低频图像进行奇异值分解，然后将水印图像嵌入到低频图像的奇异值变换域上。在水印的嵌入与提取过程中只对4帧视频图像进行小波变换，所需内存较小，并易于硬件实现。实验结果表明，本文提出的视频水印算法具有较好的隐蔽性和鲁棒性。

For the copyright protection and information hiding of digital video，a new video watermarking algorithm based on interframe wavelet transform is proposed. Firstly, video frames are divided into groups with the same frame number, and interframe wavelet transform is performed on each group with maximum levels, which gets several high frequency frames and one low frequency frame. The low frequency frame contains most of the energy of video sequence, while the high frequency frame contains little energy. Then, watermarking is embedded into the low frequency frame in the SVD field. The proposed watermarking embedding and extracting algorithm uses only four frames for performing wavelet decomposition per operation with lower memory demands. The experimental results show that this method has very good imperceptibility and good robustness.