不可分小波构造及其在图像水印中的应用

Non-separable wavelet construction and its application in image watermarking

针对张量积小波数字图像水印算法在几何攻击下鲁棒性较差的问题,提出一种将不可分小波、奇异值分解以及Arnold置乱变换相结合的数字图像水印算法。该算法首先构造了二维不可分小波滤波器组,其次对水印图像进行Arnold置乱变换,对原始图像用不可分小波进行多尺度分析,然后分别对其低频部分和置乱后的水印进行奇异值分解,最后在两者的奇异值矩阵中嵌入水印。实验结果表明,该算法能实现水印的嵌入和准确提取,并且水印的不可见性好,与张量积小波相比,该方法在对滤波、噪声、剪切、旋转、加噪的攻击下,鲁棒性较强。

In view of the poor robustness of tensor product wavelet digital image watermarking algorithm under the geometric attack, a digital image watermarking algorithm in combination with the non-separable wavelet, singular value decomposition(SVD) and Arnold scrambling transform is proposed. In the algorithm,a filter bank of two-dimensional non-separable wavelet is constructed first. And then,the watermarking image is subjected to the Arnold scrambling transform,and the non-separable wavelet is used to perform multi-scale analysis on the original image. The low-frequency part and the scrambled watermarking are subjected to the SVD. In the end,the watermarking is embedded in the singular value matrix of the two. The experimental results show that the algorithm can accurately embed and extract watermark,which is of good invisibility performance. In comparison with the tensor product wavelet,the proposed method is of better robustness under the attack of filtering,noise,shearing,rotation and noise adding.