基于SVD-Winograd快速变换的半盲水印算法

目前多数基于傅里叶变换域的方法，其抗几何攻击性能虽很好，嵌入的信息量也很大，但存在抗剪切攻击能力较差的问题，针对这一问题提出了一种基于奇异值分解与Winograd快速傅里叶变换相结合的半盲水印算法。在奇异值分解过程中，先将奇异值分解的对角矩阵与Winograd快速傅里叶变换中分解的对角阵进行矩阵相与，产生一新对角矩阵。再将数字水印嵌入到新对角矩阵中，最后通过相关运算提取水印。实验结果表明，该方法可获得较好的图像视觉效果，对剪切、噪声、旋转等攻击皆具有较好的鲁棒性。

Semi-blind watermarking algorithm based on SVD-Winograd fast transform

At present most of the methods based on Fourier transform domain have very good performance against geometric attacks and with a great amount of information embedded, but they have poor resistance to shear attack. Therefore, a semi-blind watermarking algorithm based on Singular Value Decomposition (SVD) and Winograd fast Fourier transform combination was proposed. In the SVD process, the first singular value decomposition of the diagonal matrix with the Winograd fast Fourier transform decomposition of the diagonal matrix to matrix multiplication, resulting in a new diagonal matrix. Then digital watermark was embedded into the new diagonal matrix. Finally the watermark was extracted through the relevant operations. The experimental results show that the method can obtain better image visual effects, and is of good robustness against shear, noise, rotation and other attacks.