基于奇异值分解的Contourlet域稳健性数字水印算法*

提出了一种基于奇异值分解的Contourlet域数字水印算法。对置乱后的水印图像进行奇异值分解,在Contourlet域中选取合适的方向子带,利用得到的奇异值来调制系数矩阵,然后通过逆变换获取嵌入水印后的图像。在水印提取中只需要保存原始图像的奇异值,实现了水印的近似盲提取。最后,进行了一系列的攻击实验,证明了与传统小波域水印算法相比,该算法的不可见性和鲁棒性都有了较大的提高。

Robust digital watermark algorithm in Contourlet domain based on singular value decomposition

To protect the copyright of digital media, this paper proposed a digital watermark algorithm in Contourlet domain. The original image was decomposed with Contourlet transformation(CT), the scrambled watermark image was decomposed with singular value decompositino(SVD), then, the singular values of watermark image were used to modulate the chosen Contourlet coefficient matrixs. After inverse transformation, obtained a watermarked image. In watermark detection, only the SVD vector of the original image was needed, which meant the blind detection was nearly achieved. Finally, applied a series attacks, the results show that compared with traditional watermark algorithm in wavelet domain, the proposed method achieves great impovement in both invisibility and robustness.