采用图像几何校正和分块压缩感知测量值奇异值分解的鲁棒水印技术  

Robust Digital Watermarking Algorithm Based on Image Geometric Correction and SVD of Block Compressive Sampling Measurements

作　　者：池晓芳 冯桂[1] 董晓慧[1] 

机构地区：[1]华侨大学信息科学与工程学院,福建厦门361021

出　　处：《华侨大学学报（自然科学版）》2015年第5期535-540,共6页Journal of Huaqiao University(Natural Science)

基　　金：福建省自然科学基金资助项目(2014J01242)

摘　　要：针对直接在图像压缩采样测量值上嵌入水印的方案对常规攻击和几何攻击鲁棒性差的问题,提出一种基于图像几何校正和分块压缩感知测量值奇异值分解(SVD)的鲁棒水印方案.对测量值的奇异值进行量化嵌入水印,并在水印检测前,用图像归一化技术对受到几何攻击的含水印图像进行几何校正,确保水印可以被正确提取.实验表明:该方法提高了水印对常规攻击和几何攻击的鲁棒性.It will lead to the poor robustness when the watermark is directly embedded into the compressive sampling measurements.To solve this problem,a robust digital watermarking algorithm based on image geometric correction and singular value decomposition（SVD）of block compressive sampling measurements is proposed in this paper.The watermark is embedded into the singular values of the image in compressive sampling transform domain.Meanwhile,the watermarked image under geometric attacks is corrected using the image normalization technique.After geometric correction,the watermark can be correctly extracted.Experimental results have shown that the proposed algorithm achieves the better robustness to common attacks and geometric attacks.

关 键 词：数字水印图像 分块压缩感知 奇异值分解 几何攻击 几何校正 鲁棒性 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]