基于DWT和SIFT的鲁棒图像水印算法

Robust image watermarking algorithm based on DWT and SIFT

为了提高数字图像水印的不可见性和鲁棒性,文章提出了一种基于离散小波变换(discrete wavelet transform,DWT)和尺度不变特征转换(scale-invariant feature transform,SIFT)的数字图像水印算法。该算法首先利用DWT找出原始图像中纹理较复杂区域,再在所找出的纹理复杂区域中提取出不变特征点;通过自适应确定局部特征区域,并对特征区域进行一级DWT,取其低频部分进行奇异值分解(singular value decomposition,SVD);最后对待嵌入水印图像进行Arnold置乱,把置乱后的一维水印信息进行SVD并通过加性准则嵌入到原始图像的奇异值中,以实现水印的嵌入。仿真结果表明,该算法不仅具有较好的透明性,而且对常规攻击和一般的几何攻击有较高的鲁棒性。

In order to improve the invisibility and robustness of the digital image watermarking, a digit- al image watermarking algorithm based on discrete wavelet transform（DWT） and scale-invariant fea- ture transform（SIFT） is proposed. In this algorithm, a more complex texture area of original image is firstly discovered by using DWT. Then the invariant feature points in the complex texture area discov- ered are extracted. The local feature area is determined by self-adaption and processed by one-level DWT, and the low frequency part is processed by singular value decomposition（SVD）. Finally, the watermark image to be embedded is disordered by Arnold. The one-dimensional watermark informa- tion is processed by SVD and embedded into the original singularity figure by using the additive rule. Thus the embedding of the watermark can be achieved. algorithm has superior transparency and high robustness The simulation results demonstrate that this to conventional attack and geometric attack.