一种新的基于奇异值分解的小波域盲水印*

目前，如何增强鲁棒性水印抵抗几何攻击尤其是旋转攻击的鲁棒性，仍是水印研究者研究的难点。为此，结合奇异值分解的特性和小波变换的优点，提出一种新的基于奇异值分解的小波域盲水印算法，对于常见的几何攻击具有很强的鲁棒性。首先对小波分解后的低频子带分块以提高水印的嵌入容量，再对各子块进行奇异值分解，将Arnold置乱后的水印嵌入到奇异值中，并使用量化方法增强水印的鲁棒性。实验表明，该算法对于常见的几何攻击特别是旋转攻击具有很强的鲁棒性，也能够抵抗JPEG2000压缩等信号处理。

Novel blind watermarking based on singular value decomposition and wavelet transform

At present, how to improve the robustness of robust watermarking against geometrical attacks, especially attacks of rotation, is still a difficulty of watermarking researchers. In this paper, in combination with singular value decomposition (SVD) properties and advantages of discrete wavelet transform (DWT), a novel blind watermarking based on SVD and DWT is proposed, which is robust to common geometrical attacks. In the embedding, the lowest frequency components of the source image which are transformed into DWT are divided into subblocks in order to improve watermarking capacity. Then the watermarks are embedded in the singular values of subblocks after being Arnold scrambled .And the robustness of watermarking can be improved with quantitative methods. Experimental results have shown that the proposed watermarking strategy is robust to common geometric attacks, especially rotating attack, as well as image manipulations such as JPEG2000 compression.