基于混合变换域和奇异值分解的零水印算法

摘要

为了克服传统水印算法需在载体图像中嵌入水印而导致的水印不可见性与鲁棒性之间的矛盾，提出一种DCT-DWT混合变换域和奇异值分解SVD（ Singular Value Decomposition ）相结合的数字图像零水印算法。算法首先对原始图像进行离散小波变换，然后对逼近子图进行分块离散余弦变换，在各分块中，对由特定数量的块内系数构成的矩阵进行奇异值分解。提取各分块最大奇异值组合成矩阵，二值量化后与水印结合生成注册零水印信息，保存到知识产权数据信息库IPR（ Intellectual Property Rights ）中。仿真实验结果表明，该算法不但有效避免了水印图像失真，而且能够很好地抵抗各种常见的攻击，具有很好的鲁棒性。%In order to avoid the contradiction between the invisibility and the robustness caused by traditional watermark algorithm that it needs to embed watermark into carrier image , in this paper we propose a digital image zero-watermarking algorithm which combines DCT-DWT hybrid transform domain and singular value decomposition ( SVD) .The algorithm first applies discrete wavelet transform ( DWT) on primitive image, and then applies sub-block discrete cosine transform (DCT) on approximation sub-image.In each sub-block, the singular value decomposition ( SVD) is applied to the matrix formed by within-block coefficients with specific amount .Maximal singular values in each sub-block are then extracted to compose a matrix , after the binary quantisation is executed , this matrix is combined with watermark to form the registered zero-watermark information and to be stored to data repository of intellectual property rights ( IPR ) . Results of simulation experiments demonstrate that the algorithm effectively avoids the watermark image distortion , and has rather good robustness and can well resist various common attacks .