基于SURF的小波域自适应水印算法

Adaptive watermarking algorithm based on SURF in wavelet domain

数字水印技术作为信息隐藏的分支领域,是进行数字版权保护的有效手段。针对水印嵌入强度的不确定性问题,同时为增强水印抵抗旋转攻击的能力,提出一种基于加速鲁棒特征(SURF)的小波域自适应水印算法。首先,对原始载体图像进行两级离散小波变换(DWT),提取低频子带进行奇异值分解(SVD);然后对水印图像进行奇异值分解,利用加性嵌入规则将水印图像的奇异值矩阵叠加到原始载体图像的奇异值矩阵中,再经过逆变换即可得到含水印载体图像;利用加速鲁棒特征算法对经过旋转攻击的含水印载体图像进行旋转校正。对于嵌入强度不确定性,构建适应度函数,利用麻雀搜索算法(SSA)对其寻优进而达到水印嵌入强度自适应的目的。实验结果表明,含水印的载体图像有良好的不可感知性,峰值信噪比(PSNR)在40 dB以上。所提算法对于常见的水印攻击具有较强的鲁棒性,归一化相关系数(NC)大多都在0.98以上。

As a subfield of information hiding,the digital watermarking technology is an effective means for digital copyright protection.In view of the uncertainty of watermark embedding strength and the purpose of strengthening the watermark′s resistant ability to rotation attack,an adaptive watermarking algorithm in wavelet domain based on speeded⁃up robust feature(SURF)is proposed.Two⁃stage discrete wavelet transform(DWT)is performed on the original carrier image,and low⁃frequency sub⁃band is extracted for singular value decomposition(SVD).Then the watermark image is subjected to the singular value decomposition.The singular value matrix of the watermark image is superimposed on the singular value matrix of the original carrier image by the additive embedding rule,and then the watermarked carrier image can be obtained by inverse transformation.Rotation correction of watermarked carrier image subjected to rotation attack is performed by SURF algorithm.For the uncertainty of embedding strength,a fitness function is constructed,and the sparrow search algorithm(SSA)is used to optimize it to achieve the purpose of watermark embedding strength self⁃adaptation.The experimental results show that the watermarked carrier image has good imperceptibility,and its peak signal⁃to⁃noise ratio(PSNR)is above 40 dB.The proposed algorithm is robust to common watermark attacks,and most of the normalized correlation coefficients(NC)are above 0.98.