一种支持向量回归机的音频水印算法

为了同时保证算法的不可感知性与鲁棒性，提出了一种新的支持向量回归机音频水印算法．该算法首先对音频进行小波变换，然后选取稳定的低频系数的相邻均值向量、左邻域斜率向量、右邻城斜率向量作为支持向量回归机训练模型，利用支持向量回归机训练模型嵌入和提取水印信息．仿真实验表明，算法对常规的信号处理及攻击均具有较好的鲁棒性，与经典的基于小波变换的量化算法相比，具有很好的不可感知性，另外，在水印检测时不需要原始音频，实现了水印的盲检测．

An Audio Watermarking Algorithm Based on SVR

A novel digital audio watermarking scheme based on the support vector regression （SVR） has been proposed to ensure the algorithmic intangibility and robustness. Firstly, the segment energy and the slope of nearby coefficient are selected as train sample, which are extracted from the audio segments after DWT, and then the support vector regression train model can be obtained by applying the support vector regression theory. The digital watermark can be embedded and extracted by utilizing the strong learning ability of support vector regression. The digital watermark can be correctly extracted under various attacks. Experimental results show that the proposed scheme is not only invisible but also robust against various attacks. In addition, the proposed scheme has outperforming performance over traditional watermarking systems and can extract the digital watermark without the help of the original digital audio.