基于关键熵的双树复小波域盲图像水印算法  

Key entropy-based blind image watermarking algorithm in the dual-tree complex wavelet transform domain

作　　者：刘金华[1] 佘堃[1] 

机构地区：[1]电子科技大学计算机科学与工程学院,四川成都611731

出　　处：《光电子．激光》2011年第5期757-762,共6页Journal of Optoelectronics·Laser

基　　金：国家自然科学基金资助项目(60873185;60903074)

摘　　要：设计了一种基于关键熵的盲数字图像水印算法。首先,使用尺度不变特征变换(SIFT)方法,从图像中提取特征点;其次,以特征点为中心构造局部不变圆形区域,并对其进行归一化处理;然后,选取大于图像平均熵的图像区域作为关键熵图像区域;最后,结合量化调制策略及双树复小波变换(DTCWT)技术,将水印嵌入到关键熵图像区域中。实验分析表明,本文算法对常用的图像处理攻击(如加性噪声、中值滤波J、PEG压缩等)和几何攻击(如缩放、旋转、仿射变换等)均具有较好的鲁棒性。A blind image watermarking algorithm based on key entropy using the dual-tree complex wavelet transform（DTCWT） and the scale invariant feature transform（SIFT） is proposed in this paper.Firstly,the image feature point is extracted by using the SIFT when generating local invariant circular regions in watermark embedding.Secondly,each circular region is normalized and the entropy of each circular region is calculated,whereas the circular regions which its entropy is greater than average entropy,are used as the ＂key entropy＂ image regions.Finally,the watermark sequences are embedded into the amplitude and phase of each key entropy image region by the quantization method in the DTCWT domain.Experimental results and comparisons have demonstrated the effectiveness of the proposed watermarking algorithm and it has good robustness against common image processing attacks,such as Gaussian noise,JPEG compression and median filtering,and some kinds of geometric distortions,consisting of rotation,scaling,affine transformations,etc.

关 键 词：数字图像水印 关键熵 尺度不变特征变换(SIFT) 双树复小波变换(DTCWT) 

分 类 号：TP391.41[自动化与计算机技术—计算机应用技术]