基于奇异值分解的小波域图像水印算法  

Novel Image Watermarking Algorithm in Wavelet Domain Based on SVD

作　　者：杜肖山[1] 廖述剑[1] 

机构地区：[1]太原理工大学信息工程学院,山西太原030024

出　　处：《计算机仿真》2011年第10期228-231,281,共5页Computer Simulation

摘　　要：研究优化图像水印问题,传统的基于奇异值分解(SVD)的图像水印算法中,由于水印信息均分布于整个图像,当含水印图像遭到攻击后,提取的水印图像有明显失真,容易出现黑色细线。为消除失真,改善水印提取效果,提出了一种奇异值分解的小波域图像水印算法。首先,在构造二值水印图像时,将文字信息分布于图像左下角和右上角,由于在对角线上并未分布信息,提取出水印图像直接将对角线的黑色细线通过像素灰度值调整为同底色相同的白色即可消除失真。水印嵌入过程中运用矩阵相乘将传统SVD算法中需保留的参数矩阵由3个减少为1个,简化了算法。实验结果表明,改进的算法既保持了鲁棒性和不可见性,同时也消除了失真,获得了很好的图像水印提取效果,可用于版权保护。Because the watermark information is distributed in the whole image,great distortion and black thin line exist around the main diagonal pixels while employing a conventional SVD-based method for extracting and recovering the watermark from the cover image which is attacked.In order to solve the problem,a novel image watermarking algorithm based on SVD in wavelet domain is presented in this paper.Firstly,a special binary meaningful image should be constructed and the text information is set in the lower left and upper right corner of the image,in which there is no text information around the main diagonal pixels.A good result can be obtained when we extract and recover the watermark by means of simple computing.Next,the multiplication of matrix is applied to reduce the saving parameters from 3 to 1 to make the algorithm simple.Experimental results indicate that the proposed algorithm is invisible and robust and also eliminates the distortion.Finally,a good watermark extraction is obtained and the algorithm can be employed for copyright protection.

关 键 词：图像水印 图像置乱 混沌加密 离散小波变换 奇异值分解 鲁棒性 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]