一种基于DWT_SVD的抗几何攻击彩色图像水印算法  

Color Image Watermarking Algorithm Against Geometric Attacks Based on DWT_SVD

作　　者：谢斌[1] 彭林[1] 刘珊[1] 

机构地区：[1]江西理工大学信息工程学院,江西赣州341000

出　　处：《电视技术》2015年第21期10-14,共5页Video Engineering

基　　金：国家自然科学基金项目(61363076);江西省青年科学基金项目(20122BAB211019)

摘　　要：针对传统离散小波变换(DWT)水印算法不能较好地抵抗几何攻击的缺点,提出了一种基于离散小波变换(DWT)和奇异值分解(SVD)相结合的彩色图像水印新算法。首先对原始彩色载体图像进行RGB三基色分解,然后进行离散小波变换,再选取4个子块的1/4重构成新子块,并进行二次离散小波变换,最后对其低频部分进行奇异值分解嵌入水印信息,得到嵌入水印的彩色图像。实验结果表明,该算法能够较好地抵抗诸如高斯噪声、椒盐噪声、压缩等常规攻击,并对大角度旋转、任意角度旋转、剪切加旋转等几何攻击也表现出了较强的鲁棒性,其总体性能明显优于传统的DWT水印算法。In the traditional DWT domain watermarking algorithm is not able to resist geometric attacks disadvantages, this paper puts forward a new color image watermarking algorithm based on the combination of DWT and SVD. The algorithm is first applied to the original carrier image of RGB decomposition, the decomposition after the RGB three component of the discrete wavelet transform, and then select the one forth sub block to form a new sub blocks, then take the second time discrete wavelet decomposition, choose the low frequency part in the singular value decomposition, finally the watermark information embedded into the carrier image singular value decomposition of the diagonal matrix, to obtain the watermarked image.Simulation results show that, the algorithm can effectively resist such as Gauss noise, salt and pepper noise, compression and other conventional attack, and the large angle rotation, shear and rotation and other geometric attacks also showed robust DWT watermarking, the overall performance is better than the traditional algorithm.

关 键 词：数字水印 离散小波变换 奇异值分解 鲁棒性 

分 类 号：TP391.4[自动化与计算机技术—计算机应用技术]