针对数字视频的版权保护,提出了一种以二值图像作为水印和基于三维小波变换及HVS特性的视频水印算法.首先对水印图像进行混沌加密和Arnold置乱处理,利用基于方向经验模式分解的视频场景分割技术将视频信号切分为不同的场景,通过密钥从中选取若干场景进行三级三维小波变换.再借助HVS对图像纹理区域、运动区域的掩蔽特性,使水印自适应被嵌入到低频子带的重要系数中.实验结果表明,文中的算法具有较好的不可见性,对于诸如Gaussian噪声、丢帧、帧平均和MPEG压缩等攻击具有很好的鲁棒性.
For the copyright protection of digital video,a video watermarking algorithm is proposed by using the visible binary image as watermarking based on three-dimensional discrete wavelet transform(3D DWT)and HVS.First,the watermarking is preprocessed by using chaotic encryption and Arnold scrambing,and the algorithm partitions the video sequence into different scenes by using video scene segmentation method based on directional empirical mode decomposition(DEMD),some of which are selected with a key to perform 3 levels 3D DWT.Second,the watermarking is adaptively embedded into the low frequency wavelet significant coefficients by using the HVS masking characteristics of texture area and motion area.Experimental results show that the proposed algorithm is well invisible and robust to attack such as Gaussian noise,frame loss,frame average and MPEG compression.