基于小波系数树权值关系的盲水印算法  

A Blind Watermarking Algorithm Based on the Relationship between Weight Values of Wavelet Coefficient Trees

作　　者：刘泽平[1] 易叶青[2] 

机构地区：[1]湖南人文科技学院,娄底417000 [2]湖南大学软件学院,长沙410082

出　　处：《微计算机信息》2010年第32期212-213,211,共3页Control & Automation

基　　金：基金申请人:易叶青;项目名称:无线传感器网络中不依赖MAC认证的虚假数据过滤算法研究;基金颁发部门:湖南省自然科学基金(09JJ6097);基金申请人:刘云如;项目名称:数字图像中的多水印协作认证机制与算法;基金颁发部门:湖南省教育厅科研项目(07C386);基金申请人:刘泽平;项目名称:基于小波变换的鲁棒图像数字盲水印算法研究;基金颁发部门:校级青年基金项目(2009QN02)

摘　　要：提出一种新的基于小波系数树权值关系的盲水印算法,算法对原始图像进行三级小波变换,构成以低频子带为根节点的小波系数树带,通过比较小波系数树的权值大小关系,结合人类视觉特性确定水印嵌入强度,将一位水印信号嵌入到从低频到高频树系数中。在水印提取时无需参考原始图像,实现了全盲检测。实验结果表明:对抵抗常见的叠加噪声、图像裁剪、JPEG压缩等攻击,水印具有较好的鲁棒性和不可见性。A blind watermarking algorithm based on the relationship between weight values of wavelet coefficient trees is proposed.Firstly,decompose the original image using three-layer wavelet transform,and construct a wavelet trees with its low frequency coefficient as root.Secondly,compareing with the size relationship between weight values of wavelet coefficient trees,and combining the characteristics of human visual system to control embedded watermark strength,embed each watermark bit into the coefficient trees which is from low frequency to high frequency.During the period of watermark extraction,no referential original image is needed.Thus realized complete blind detection.The experimental results indicate that the watermark have good invisibility and robustness in resisting the attacks such as superimpose noise,image cropping,JPEG compressing,ect.

关 键 词：小波变换 人类视觉系统 小波系数树 盲检测 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]