摘要： 为改善视频水印抗H.264压缩的性能，提出一种基于光流场的视频水印算法。根据视频图像中各点的光流速度，计算其帧图片的总光流，再根据总光流的差值来选定视频关键帧，在这些关键帧中利用DCT分块变换来嵌入水印信息。实验结果表明，该算法能有效对抗H.264的3种压缩模式，满足了水印的鲁棒性和视频高压缩率的双重要求。

关键词: 光流场, 视频水印, H.264压缩, 鲁棒性

Abstract: In order to improve the performance of the video watermark resist H.264 compression, this paper proposes a video watermarking algorithm based on Image Optical Flow(IOF). According to the optical flow rate of the points in video images, the total optical flow in the frame is calculated. The key video’s frames are selected on the basis of the dispersion of the total optical flow. And with the DCT block transform, the watermark information is embedded in those frames. Experimental results show that the algorithm can effectively antagonize the three compression modes of H.264 and ensure the robustness of the watermark and multiple requirements of the high image quality as well as the high compression ratio.

Key words: optical flow, video watermark, H.264 compression, robust

中图分类号: 

• TP391.41