基于分块的多级DCT和SVD的鲁棒水印算法  

A robust watermarking algorithm based on block-divided multi-level DCT and SVD

作　　者：王琳娟[1] 李君霞[2] 

机构地区：[1]山西农业大学文理学院,山西太谷030801 [2]中国电信集团山西省分公司信息化部

出　　处：《计算机时代》2017年第9期51-55,共5页Computer Era

基　　金：校科技创新基金(201209)

摘　　要：通过仿真实验和K-S假设检验验证了离散余弦变换(DCT)具有"能量集中"的特点和奇异值分解(SVD)具有抗旋转,抗镜像,抗转置,抗剪切的特点。为了充分利用这些特点,针对大多算法在嵌入水印时需要调制嵌入的问题,提出一种基于分块的多级DCT和SVD的鲁棒性水印算法,该算法在嵌入水印时进行多级DCT变换和SVD分解、采用直接替换方式嵌入,设置多重密钥以提高算法的鲁棒性。仿真实验及分析表明,该算法抗噪、抗压缩和抗几何攻击的能力都较强,是一种强鲁棒性算法。By simulation and Kolmogorov-Smimov （K-S） hypothesis test, it is confirmed that the discrete cosine transform （DCT） has the characteristics of energy concentration and the singular value decomposition （SVD） has the characteristics of anti-rotation, anti-mirror, anti-transposition, anti-cut etc. To take full advantage of all characteristics of DCT and SVD and to overcome the fault that most algorithms need embedding watermarks by modulation technique, a robust watermarking algorithm based on block-divided multi-level DCT and SVD is proposed. In order to improve the robustness of the algorithm, multi-level DCT and SVD, and the direct replacement method are used to embed the watermarks. In addition, multiple keys are set for extracting the watermarks. The simulation results show that the algorithm has a stronger robustness against noises, compress and geometric attacks, and it＇s a strong robust algorithm.

关 键 词：数字水印 SVD 多级DCT 分块 

分 类 号：TP309.7[自动化与计算机技术—计算机系统结构]