Nowadays many watermarking algorithms have been proposed and implemented for multimedia watermarking. Compared to the image watermarking and audio watermarking, the data of video watermarking is larger and the processing is more complex. In this paper, the FFMPEG platform is used which can greatly reducing the complex programming. In order to improve invisibility and robustness of the watermarking, a key frame of the MPEG2 video is extracted and the discrete cosine transform(DCT) is implemented. We use the Itti’s visual saliency model to distinguish visual salient regions and after scrambling binary watermark, then the DCQIM is adopted to embed the watermark on visual nonsalient regions to ensure that the embedding watermark on the video achieve the minimum impact. The results of the experiment show that the algorithm has good invisibility and has good robustness to common attacks.