基于分块奇异值分解的小波域水印算法  

An Algorithm of Digital Watermarking Based on Block Singular Value Decomposition in Wavelet Domain

作　　者：龚劬[1] 马素春[1] 

机构地区：[1]重庆大学数理学院,重庆400030

出　　处：《计算机仿真》2009年第5期138-141,共4页Computer Simulation

摘　　要：对于一般有意义灰度水印,如何在不可见性下增大水印嵌入容量,增强鲁棒性仍是水印研究学者研究的重要内容。根据奇异值分解的特性,结合小波变换与人类视觉系统的某些特性接近的良好特点,提出了一种基于分块奇异值分解的小波域鲁棒水印算法。算法首先将灰度水印信息分为重要信息和次要信息两部分,然后在小波域对宿主图像小波分解后的高频和低频系数分别进行分块奇异值分解,再把经过置乱变换的重要信息部分和次要信息部分分块DCT变换,将变换后的水印信息分别嵌入低频和高频系数分块奇异值分解的奇异值中。实验结果与分析表明:水印算法能嵌入大容量水印,且能抵抗大多数图像攻击,是一种可行的算法。For meaningful gray level watermarking scheme, how to improve watermarking capacity and its robustness under guarantee of invisibility is still an important part for watermarking researchers. As the singular value decomposition is a special matrix transform with good properties, and the wavelet transform is excellent local time - frequence analysis and muhianalysis, this paper proposes a robust digital watermarking algorithm based on block singular value decomposition in wavelet domain. The gray level image is divided into important and unimportant parts, each part is decomposed with blocked - DCT transform after being Arnold scrambled. With the human system （HVS） in the domain of DWT, the original image is decomposed with DWT into several subbands, and block - SVD is applied to transform the subbands. Then the important message is embedded in the SVD of low freguency suband, and the unimportant message is inserted into the SVD of high frequency subband. Experiment and analysis show that the algorithm has large embedding capacity and is strongly robust for the common image processing techniques.

关 键 词：小波变换 奇异值分解 数字水印 鲁棒性 

分 类 号：TP393.08[自动化与计算机技术—计算机应用技术]