ASIFT与SVD相结合的Contourlet域的抗几何攻击视频双水印算法

Video Dual Watermarking Algorithm Based on Moving Object Detection in Contourlet Domain

为平衡数字水印技术中的鲁棒性与透明性,提出基于仿射尺度不变特征变换(ASIFT)与奇异值分解(SVD)结合的Contourlet域视频双水印算法。首先,利用卡尔曼滤波器将视频帧内的运动物体进行分离获取运动宏块,对运动宏块进行Contourlet变换;其次,利用奇异值分解对图像的信号攻击存在较高的稳定性和低频子带系数直方图在旋转、缩放等几何攻击具有较强的鲁棒性,自适应地将混沌水印序列分别嵌入到高频子带奇异值矩阵和低频子带系数中;最后,以ASIFT的尺度和方向不变性作为触发器,用于判断嵌有水印的视频遭受的攻击类型。本文算法能保证嵌入水印的视频帧的峰值信噪比值(PSNR)在70 dB以上,水印的嵌入量达到240×360 bit。实验结果表明本文的算法能够保证水印的不可见性的同时针对几何攻击和信号攻击都具有较强的鲁棒性。

In order to balance the robustness and transparency in digital watermarking,dual watermarking algorithm based on affine scaling invariant feature transform( ASIFT) and singular value decomposition( SVD) was proposed. Firstly,Kalman filter was used to separate the moving objects for each frame. The moving objects were processed by the Contourlet transform,the high frequency subband with the highest energy was selected. The singular value matrix has the stability of signal attacks. The watermark sequences are adaptively embedded into the high frequency sub-band of singular value matrix. Then,the low frequency sub-band coefficient histogram is robustness under geometric attacks such as rotation and scaling,and the watermark signal was embedded in the adjacent coefficient level of the low frequency sub-band histogram. Finally,the ASIFT scale invariance and direction invariance was used as the trigger to judge whether the video frame is subjected to the geometric attack or not.The algorithm can guarantee that the Peak Signal to Noise Ratio( PSNR) value of the video frame is more than70 dB,the embedded amount of the watermark is 240×360 bits.Experimental results show that this algorithm can guarantee the invisibility of watermarking and has strong robustness against both geometric and signal attacks.