基于双树四元数小波变换与奇异值分解的数字图像水印算法

Digital Image Watermarking Algorithm Based on Dual Tree Quaternion Wavelet Transform and Singular Value Decomposition

本文提出了一种新的基于双树四元数小波(DTQWT)和奇异值分解(SVD)的非盲数字图像水印算法,该算法对几何攻击和普通图像篡改具有较强的鲁棒性。在把双树四元数小波应用于原始载体图像后,我们将奇异值分解应用于它的低频幅度子带LL,然后对原始水印图像进行Arnold置乱,利用置乱后的水印图像修正载体图像的奇异值。该算法的优点是对大多数常见的攻击具有较强的鲁棒性。分析和实验结果表明,与基于RDWT-SVD的方法相比,本文算法具有更好的性能。

In this paper, a new non-blind digital image watermarking algorithm based on dual-tree quaternion wavelet (DTQWT) and singular value decomposition (SVD) is proposed. The algorithm is robust to geometric attacks and ordinary image manipulation. After applying the dual-tree quaternion wavelet to the original cover image, we apply the singular value decomposition (SVD) to its low frequency amplitude sub-band LL, then scramble the original watermark image with Arnold transform, and modify the singular value of the cover image using the scrambled watermark image. The advantage of the proposed algorithm is that it is robust to most common attacks. Analysis and experimental results show that, compared with the method based on RDWT-SVD, the proposed algorithm shows better performance.