Abstract: The existing digital image watermarking methods present insufficient affine tolerance.The main reason is that affine attacks don’t erase the watermark information，but result in the synchronization errors between watermark detection and embedding watermark procedures.To improve affine attacks tolerance of digital image watermarking，this paper proposes a novel digital watermark method based on image moments and Fourier-Mellion transform.In the proposed method，image centric moments is utilized to implement translation invariance by translation normalization，and then Fourier-Mellion Transform is employed to implement rotation and scaling invariance.The experimental results prove that the proposed method can get better visual effect，it is also shown that this method is robust to affine attacks.

Key words: image moments, digital watermarking, Fourier-Mellin transform

摘要： 现有的数字水印技术大多难以抵抗几何变换类攻击，如旋转、平移和尺度变换等，特别是仿射变换（旋转、平移和尺度变换混合的多种攻击），其中一个最主要的原因是：几何变换虽然并未去除图像中的水印信息，但却使水印的检测与嵌入之间失去同步，从而导致水印检测的失效。针对以上原因提出了一种基于图像矩和傅里叶梅林变换抗仿射不变性水印算法。对于平移造成的几何攻击，在水印检测之前利用原始图像的几何矩估计水印图像所经过的几何变换，根据估计的参数对水印图像进行相应的校正和水印检测。对于旋转和尺度变换造成几何攻击，通过对数极坐标系能将笛卡尔坐标系中的旋转和缩放变换为平移的性质，结合Fourier变换来抵抗旋转和缩放攻击。实验结果证明，该方法可以获得良好的图像视觉效果，同时对于几何攻击具有很好的鲁棒性。

关键词: 图像矩, 数字水印, 傅里叶梅林变换