一种基于多小波域与DCT的数字盲水印算法  

Blind Watermarking Algorithm Based on Multi-wavelet Domain and DCT

作　　者：赵雪章[1] 席运江[2] 

机构地区：[1]佛山职业技术学院电子信息工程系,广东佛山528237 [2]华南理工大学工商管理系,广东广州510000

出　　处：《计算机仿真》2011年第3期126-129,共4页Computer Simulation

摘　　要：研究一种新的数字图像盲水印算法。针对目前基于的离散小波变换算法和DCT变换算法中,对噪音和几何变形攻击抵抗力弱,需要原始水印的缺点,提出了一种利用多小波域和离散余弦变换相结合的数字盲水印算法。结合离散多小波变换的多分辨率特性和离散余弦变换所固有的特征,首先对原灰度图像DMWT变换,而后对低频系数作分块DCT变换,再将经过Arnold预处理的二值图像水印信息分别嵌入到经过DCT变换的较好的块中。提取水印时不需要原始图像的参与,实现了盲水印提取。实验结果表明,嵌入算法容易实现,并增强了水印的稳健性和不可见性,算法对锐化J、PEG压缩、剪切和噪声等常见的图像攻击有较好的鲁棒性。The paper proposes a new digital image blind watermarking algorithm.To solve the problems that current algorithms based on discrete wavelet transform and DCT transform have weak resistance on noise and geometric distortion attacks so that have to use original watermark.The paper suggests a new digital image blind watermarking algorithm based on the combination of multi-wavelet domain algorithm and discrete cosine transform algorithm.Combining the multi-resolution characteristics of wavelet transform and the inherent characteristics of discrete cosine transform,firstly DMWT transform is implemented for original gray image,then DCT-block transform for low-frequency coefficients,and binary image watermarks preprocessed through Arnold are embedded in the blocks with better DCT transform results.The presented algorithm can extract blind watermark without using original image.Experiments show that the presented algorithm has many advantages： easy to be carried out,enhancement of robustness and non-visibility of the watermark,and better robustness for common image attacks such as algorithm sharpening,JPEG compression,shearing,noises and so on.

关 键 词：盲水印算法 离散多小波变换 离散余弦变换 鲁棒性 

分 类 号：TP309.2[自动化与计算机技术—计算机系统结构]