摘要： 提出了一种基于Watson 视觉模型的自适应视频水印，Watson 视觉模型用于控制水印的嵌入强度。借助于该视觉模型，很容易权衡水印鲁棒性和感知透明性的关系。经过实验证实，该水印系统对常用的视频攻击(MPEG 压缩、帧丢失、帧剪切、帧重排)是鲁棒的，同时具有感知透明性和实时性等特性

关键词: 视觉模型；视频水印；位错率；鲁棒性

Abstract: This paper proposes a video watermarking based on the Watson visual model, which is introduced to control the watermark embedding strength. Assisted by the Watson visual model, it is convenient for balancing the relationship between watermark robustness and visual transparency. Experimental results indicate this scheme is robust to common video attacks, including MPEG compression, frame drop, frame crop and frame re-arrange. Moreover, the scheme has the property of perceptual transparency

Key words: Visual model; Video watermarking; Bit error rate; Robustness