基于原始图像矩的仿射不变性水印算法

Affine Invariant Watermarking TransformBased on Geometric Moment

现有水印技术大都难以抵抗几何变换类攻击,如旋转、平移和尺度变换等,即使微小幅度的图像旋转或尺度变换都可能导致水印检测过程失败.作者提出一种基于原始图像矩的仿射不变性水印算法.在水印检测前利用原始图像一个或多个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校正和水印检测,该方法可在任意作用域中实现.给出一种在DCT域中的实现方法,该几何攻击由通用的水印攻击软件Stir mark产生,水印检测过程不需要原始图像.实验结果表明,该方法对于几何攻击具有很好的鲁棒性.

Many proposed image watermarking techniques are sensitive to geometric distortions, such as rotation, translation and scaling. Rotation and scaling of slight amount can make the detector disabled unless the corrupted watermarked image can be rescaled and rotated back to its original size and original orientation. In this paper we propose a new affine invariant watermarking transform scheme based on geometric moment. The geometric distortions of watermarked image are estimated by exploiting the geometric moments of original image,so that the corrupted watermarked image can be corrected and the watermark can be detected correctly. The scheme of moment estimation is available to various watermark systems as a preprocess method before watermark extraction. A DCTdomain watermarking method is given as an example. Experimental results show that the proposed method is robust to geometric distortions,even to Strimark attacks.