一种新的基于奇异值分解的图像数字水印算法  

A New Image Digital Watermarking Algorithm Based on SVD

作　　者：徐慕蓉[1] 樊锁海[1] 

机构地区：[1]暨南大学信息科学技术学院,广东广州510632

出　　处：《计算机仿真》2011年第5期291-294,341,共5页Computer Simulation

基　　金：国家自然科学基金(10671076;11071089);中央高校基本科研业务费专项资金(21609602);广东省自然科学基金(10151063201000005)

摘　　要：图像数字水印优化,增强了水印的抗攻击能力。在一类基于奇异值分解的数字水印算法中,若将水印嵌入图像矩阵的最大奇异值,降低了图像的鲁棒性。为提高图像和水印的鲁棒性,根据奇异值分解的特性,结合Aronld变换的分散攻击和人类视觉系统寻找最优嵌入强度的良好性能,提出了一种新的自适应图像数字水印算法。首先将图像分块,然后计算得到每块图像的最大奇异值,再将经Arnold变换后的水印结合通过人类视觉系统技术得到的水印强度嵌入最大奇异值,从而得到带水印的图像。通过仿真分析比较表明,在一定程度上提高了图像的视觉质量和水印的鲁棒性,是较有效的一种算法。In some singular value decomposition-based methods,watermark is embedded in the greatest singular value of image matrix,which enhances the ability to resist attacks of wartemark,but lowers image robustness.In order to improve the robustness of image and watermark,a new adaptive image digital watermarking is proposed,which is based on some special properties of SVD and the satisfactory performance that Arnold scrambling can disperse attacks and HVS would find the best embedding strength.First,the image is blocked and then every block is calculated to get the greatest singular value.After transformed by Arnold,the watermark is embedded in this greatest singular value combined with the watermark embedding strength of the block,which is obtained by HVS.Thus,we got the image with a watermark.Compared with the algorithm proposed by Xiao and Wang,the experimental results demonstrate that this algorithm improves the image quality and watermark robustness in a certain extent,which is a more effective algorithm.

关 键 词：数字水印 奇异值分解 图像置乱 人类视觉系统 

分 类 号：TP309[自动化与计算机技术—计算机系统结构]