基于双混沌系统和SVM的鲁棒性数字水印算法

Dual Chaotic Systems and SVM Based Robust Digital Watermarking Algorithm

支持向量机(SVM)由于其良好的分类和回归特性,在信号处理等领域有较好的应用.该文将Arnold变换和SVM回归算法结合应用在版权保护的数字水印算法中.首先将宿主图像划分为大小相等的块,用Arnold变换选取SVM的训练图像块(TB)和水印的嵌入图像块(EB),然后将Logistic混沌系统调制过的水印图像通过修改EB中心像素值将其嵌入.实验结果显示,该算法水印的具有较高的安全性和良好的视觉特性,而且可以有效抵抗噪声,剪切,JPEG等常规的攻击.

There are many applications of support vector machine （SVM） in signal field for its better classifying and regression performance. A new robust digital watermarking algorithm was presented in this paper. Firstly, host image was divided by the same size blocks, and training blocks and embedded blocks were selected by Arnold transform. Then the watermark sequences modulated by Logistic system were embedded by modifying the center pixel value in the embedded blocks. The simulation results show that the proposed algorithm not only has the high security and good visual quality, but also can resist many attacks.