基于复Contourlet域的支持向量机和Krawtchouk矩的双水印算法

为了进一步提高基于支持向量机(SVM)水印算法的 鲁棒性,提出了一种 基于复Contourlet域的SVM和Krawtchouk矩的双水印算法。首先从RGB宿主 图像中提取B 分量和G分量,并且充分利用Krawtchouk矩不变量的平移、旋转、缩放不变性 和Krawtchouk矩良好 的局部重构特性,计算B分量图像的Krawtchouk低阶矩不变量 ,由此构造鲁棒水印；然后对G分量图 像进行两级复Contourlet分解,在其低频分量中,利用SVM建立图像尺度内的 局部相关性训练模型, 并根据预测结果自适应地实现数字水印图像的嵌入和提取。大量实验结果表明,本文算法不 仅具有较好的 不可感知性,而且对中值滤波、加性噪声和JPEG压缩之类的常规图像处理,以及缩放、旋转 和剪切等几 何攻击,均具有较好的鲁棒性能,其性能优于基于小波域的SVM和基于Contourlet域的SVM水 印算法。

Double watermarking algorithm based on support vector machine in complex Contourlet domain and Krawtchouk moment

In order to further improve the robustness of watermarking algorithm b ased on support vector machine (SVM),a double watermarking algorithm based on SV M in complex Conto urlet domain and Krawtchouk moment is proposed in this paper.Firstly,the blue component and green componen t are extracted from the RGB host image.The algorithm makes full use of the invariance of Krawtchouk moment invariants to translation, rotation and scaling,and the better local reconstruction characteristics of Kra wtchouk moment.The lower order Krawtchouk moment invariants of the blue component are calculated for constructi on of robust watermarking. Then the green component image is decomposed by two-level complex Contourlet tr ansform.In the low-frequency components,the local correlation training model of image in the same scale is e stablished by using support vector machine.The watermarking image is adaptively embedded and extracted according t o the prediction results of the established model.A large number of experimental results show that the proposed algorithm is not only invisible but also robust against the common image processing,such as median filtering,Ga ussian noise and JPEG compression,and against some kinds of geometric attacks,including rotation,sca ling and clipping.It outperforms the image watermarking algorithm based on support vector machine in wavelet doma in or Contourlet domain.