中文摘要：

针对MPEG-2的视频特征，提出了一种视频水印算法。该算法利用GOP中的I帧和P帧实现定位，并在多个B帧嵌入水印信息。在确定嵌入强度时，利用人类视觉特性自适应的方法嵌入阈值。通过仿真测试，该算法具有较好的透明性和鲁棒性。

英文摘要：

According to MPEG-2 video features, a video watermarking algorithm is proposed. The positioning is achieved by using I-frame and P-frame of GOP, and the watermark information is embedded in a plurality of B frames. In determining the embedding strength, adaptive threshold is embedded by using human visual characteristics method. The simulation tests showed that the algorithm has good transparency and robustness.