一种改进的基于奇异值分解的数字水印算法

An Improved Digital Watermarking Algorithm Based on Singular Value Decomposition

针对载体图像作离散小波变换和奇异值分解,选取奇异向量矩阵的部分系数进行修改,以嵌入水印的算法,论文分析算法采用离散小波逆变换造成图像失真的原因,去除对图像的离散小波变换,直接在奇异向量矩阵中嵌入水印,理论分析表明,改进后的算法鲁棒性更强,图像失真度降低。考虑到奇异值分解的奇异向量矩阵是单位正交矩阵,在修改系数实现水印嵌入时注意对向量的单位性约束,能使得其它系数变化幅度收窄,图像整体变化差异减小。实验表明,论文提出的改进算法得到的含水印载体图具有较好的鲁棒性,且在经受轻微扰动后提取的水印质量较好。

In view of the carrier image discrete wavelet transform and singular value decomposition,part of coefficient of singular vector matrix is chosen to modify embedding watermarking algorithm,analysis algorithm of discrete wavelet inverse transformation of this step the image distortion caused by reason,and through experimental verification for many times,discrete wavelet transform,image do directly in the experimental scheme of embedding watermark the singular vector matrix algorithm and stronger robustness,image distortion degree is reduced.At the same time,considering that the singular vector matrix of singular value decomposition is the unit orthogonal matrix,when the watermark is embedded by modifying the coefficients,the unit constraint of the vector should be paid attention,so that the variation range of other coefficients can be narrowed and the overall variation difference of the image will be reduced.The experimental results show that the watermarks obtained by this algorithm have good robustness and the quality of watermarks extracted after slight disturbance is good.