基于压缩传感的分块DCT域灰度图像水印算法  

A Gray Image Watermark Algorithm based on Compressed Sensing and DCT

作　　者：詹旭[1] 罗毅[1] 陈昌忠[1] 梁小晓[2] 

机构地区：[1]四川理工学院自动化与电子信息学院,四川自贡643000 [2]四川理工学院计算机学院,四川自贡643000

出　　处：《通信技术》2015年第1期90-96,共7页Communications Technology

基　　金：四川省教育厅项目(No.14ZB0223);国家自然科学基金项目(No.61178068);四川省高校重点实验室项目(No.013WYJ03)~~

摘　　要：文中研究了压缩传感方法和图像水印算法,提出了一种基于压缩传感的分块余弦变换域灰度图像水印算法。该算法首先将原始二值水印图像采用压缩传感方法对其观测,从而得到观测矩阵,并将观测矩阵进行arnold置乱(阿诺德置乱)后生成待嵌入的水印图像;在嵌入水印阶段,首先将载体灰度图像进行了分块,分块的大小取决于水印图像的大小,然后将每个小分块进行二维余弦变换,计算其每个小分块中心像素的邻居(上下左右)像素均值,比较均值与对应水印像素位大小,以确定其嵌入数据,从而实现水印嵌入。为验证此算法的鲁棒性,我们进行了实验,实验表明该算法能承受一定的裁剪攻击。This paper studies the compressed sensing algorithm and DCT（Discrete Cosine Transform） algorithm, then proposes a gray image watermark algorithm based on compressed sensing and DCT. Firstly, the original binary watermark image is observed by compressed sensing method to obtain observation matrix. The observation matrix is encrypted by Arnold scrambling transformation to generate the watermark image. Then, the original gray image is divided into small image blocks, and its size depends on the watermark image＇s size. Next, the small image blocks are transformed by DCT and the upper and lower mean pixels of the center pixel is calculated. The embedded data is confirmed through comparing the mean and the related watermark pixel size and thus watermark embedding is achieved. Experiments are made to test the robustness of the algorithm, and the results show that this algorithm can effectively resist cutting attacks.

关 键 词：压缩传感 余弦变换 阿诺德 

分 类 号：TN919.8[电子电信—通信与信息系统]