基于曲率扩散模型的可见数字图像水印攻击算法研究

Research on Visible Digital Watermarking Attack Algorithm Based on the Curvature Diffusion Model

基于曲率扩散模型(CDD)算法是图像修复的重要算法之一,主要用来修复有划痕的旧照片和被文字覆盖的图像.文章将该算法引入可见数字水印攻击算法中,利用p-Laplace算子的非线性各向异性扩散的性能来填充图像受损区域即可见水印处.实验结果表明,CDD修复算法能有效的除去载体图像中的水印图像,并且在可见水印区域中能较好的修复出载体信息.

The algorithm based on the curvature-driven diffusions model （CDD） is mainly used to revert cracks in photographs and remove texts from photos. This article introduce the visible digital watermark attack algorithm and use nonlinear anisotropic diffusion properties of p-Laplace operators to fill a damaged area of the image that is visible watermarking. Experimental results show that the CDD repairing algorithm can effectively eliminate the watermarking image of carrier image.