抗几何失真的局部数字水印算法

抗几何攻击一直是数字水印中的难点问题,为了更好地解决这个问题,基于图像的几何不变域,提出了一种抗几何失真的局部数字水印算法.算法利用显著性检测器检测到的具有平移、缩放和旋转不变性特性的圆形显著区域来嵌入水印.如何利用显著区域构造不变区域以获得平移、缩放及旋转不变性,以及如何从不变区域中筛选出适合嵌水印的区域,为要解决的重点问题,首先将水印进行几何变换来匹配各个水印嵌入区域的形状,再不可感知地嵌入到各个水印嵌入区域中.还着重讨论了如何利用显著区域获得平移、缩放及旋转不变性.实验表明该算法能很好地抵抗一般的信号处理攻击及旋转、缩放、裁剪和线性变换等几何攻击.

Localized Image Watermarking against Geometrical Distortions

How to resist geometrical attacks is a traditional difficulty in digital watermarking field. To resolve it better, a localized image watermarking algorithm against geometrical attacks based on the invariant regions in the image is proposed in this paper. Salient regions detected by Saliency detector are used to synchronize the watermark. How to use the salient regions to construct the invariant regions for attaining the translation, scaling and rotation invariance and how to select the embedding regions out of the invariant regions are key problems to be resolved in the paper. The watermark is first geometrically transformed to match the shape of each embedding region and then imperceptibly embedded into corresponding embedding regions. The watermark detection process is blindly done based on the normalized correlation. Experiments results show that the proposed method is robust against common signal processing attacks and geometrical attacks including rotation, scaling, cropping and linear transformation.