基于傅里叶-梅林矩的抗几何攻击图像水印算法

如何有效抵抗几何攻击是数字图像水印研究领域的热点问题之一，一个微弱不可觉察的几何攻击就可能使绝大多数水印算法失效。以不变矩理论为基础，提出了一种基于正交傅里叶-梅林矩的可有效抵抗几何攻击的图像水印新算法。结合傅里叶-梅林矩的几何不变特性，计算出原始图像的傅里叶-梅林矩；根据稳定矩的选取规则选取部分稳定的傅里叶-梅林矩，采用量化调制策略将水印信息嵌入到所选矩的幅值中；将傅里叶-梅林矩修改前后的重构差值图像叠加到原始载体图像中，得到含水印图像。仿真实验表明，该算法不仅具有较好的不可感知性，而且对常规信号处理和几何攻击均具有较好的鲁棒性。

Robust method of digital image watermarking based on Fourier-Mellin moments

One of the hottest issues in the research areas of digital image watermarking is how to effectively resist geometric distortions. An imperceptible geometric attack can even make the vast majority of the watermarking algorithms become invalid. On the basis of the theory of invariant moments and Fourier-Mellin moments, the paper proposes a new method on image watermarking, tending to resist geometric distortions more effectively. Following the principle of invariance to geometry, the characteristics of Fourier-Mellin moments, the Fourier-Mellin moments of the original image are calculated. A part of stabilized Fourier-Mellin moments are selected according to the rules of stable moments, and the watermark information is embedded into the amplitude by quantitative modulation strategy. It reconstructs the different images before and after modification, and adds them to the original carrier image, the new watermarked image will be gained. Simulation results show that the proposed method not only has the nature of better imperceptibility, but also has an excellent robustness to the conventional signal processing and geometric distortions.