一种基于边缘检测技术的DCT域非嵌入式认证水印

根据DCT系数的特性及边缘检测技术,提出了一种依赖于可信第三方的、可有效抵抗常规图像处理攻击的非嵌入式认证水印.算法通过对图像各DCT分块中的直流系数进行计算来获取图像的边缘属性,并利用该属性生成含有水印信息的用于版权保护的认证码.传统的水印算法需要修改载体图像的内容,会损坏图像数据,而非嵌入式认证水印算法利用认证码避免了水印信息对图像数据造成破坏,保证了载体图像的质量.仿真实验结果表明,与常规的DCT域水印嵌入算法相比,该算法对普通的图像处理操作(如锐化、噪声、反色处理等)具有更好的鲁棒性,能有效抵御JPEG压缩攻击,对几何攻击(如旋转、裁剪等)也具有一定的抵抗能力.

A Non-embedded DCT-based Watermarking Using Edge Detection Technology for Authentication

In this paper,we propose a non-embedded watermarking based on a trusted third party,coefficients features of DCT and edge detection.The algorithm first calculates the edge information of the images by DCT coefficients and then computes authentication codes with the edge information.The non-embedded watermarking technology avoids the damage of the original image,which is caused by traditional watermarking.The experiments have shown that,compared with conventional DCT-based watermarking algorithm,the algorithm proposed in this paper is much more efficient in resisting general image processing operations(e.g.sharpening,noising,invert,etc),JPEG compression,and geometric attacks(e.g.rotation,cropping,etc).