一种基于正交分解的离散小波域水印方案

A WATERMARKING ALGORITHM BASED ON QR-DECOMPOSITION IN DWT DOMAIN

提出一种基于QR分解的小波域图像水印算法,是将水印信息矩阵做QR分解并嵌入到做离散小波变换后的载体图像的两个特殊的区域。首先将原始图像进行归一化处理后做离散小波变换,并计算出嵌入区域的嵌入强度;再将水印按照密钥进行置乱;然后再对置乱后的信息做正交分解,得到两个正交矩阵;分别将这两个正交矩阵使用加性规则自适应地嵌入到小波变换域的低频和对角方向上;最后小波逆变换得到含水印的图像。实验结果证明,本算法具有较好的鲁棒性、安全性和不可见性,特别是抗几何攻击效果尤其明显。

A QR-decomposition-based watermarking algorithm is developed in wayelet domain for images. The watermarking information matrices are QR-decomposed and be embedded into two special domains of the carrier image of which the discrete wavelet transform （DWT） has been done. Firstly, the host image is transformed into discrete wavelet transform domain after being normalization processed, and the embedding intensity of embedding domain is calculated ; Secondly, according to the first key, the watermarking image is transformed for scrambling; Then, QR-decomposition is implemented on the scrambled information to obtain two orthogonal matrices, and these two matrices are separately embedded into low frequency and diagonal directions of DWT domain adaptively according to adder rules; Finally, the marked image is gained through inversed discrete wavelet transform. Experimental results show that our scheme makes a better tradeoff between imperceptibility and robustness, and is secure. In particular, the effect of anti-geometric attack is evident.