基于听觉模型的自适应水印算法设计

针对音频数字水印算法抗攻击的问题，深入研究了基于人耳听觉模型的音频数字水印算法．给出了典型的基于人耳听觉模型的音频水印算法流程．提出了基于自适应量化步长的音频数字水印算法，详细分析了量化步长的自适应选取原理及计算公式，使得量化步长的选取能够动态的适应音频载体文件数据能量的变化而发生改变，提高了音频数字水印算法的抗攻击能力．实验测试结果表明，相对于固定量化步长的音频数字水印算法而言，在音频文件进行压缩攻击和低通滤波等攻击手段时，所提取的音频数字水印文件的相关系数能够达到90％以上，表现出良好的抗攻击能力．

Auto-adapted Watermark Algorithm Design based on HAS Model

An audio digital watermark algorithm based on HAS Model has been deeply studied to the issue of anti-attack of audio digital watermark, and a typical algorithm flow has also been proposed. Then the algorithm based on adaptive quantization step has also been researched. After analyzed the adaptive selection principles and computing formula of quantization step, its selection can dynamically adapt to the changes of data energy in the audio files so as to improve the anti-attack level of the watermark. According to the tests, it can be learned that the proposed method shows a better anti-attack ability compared to the fixed quantization step method, whose correlation coefficient can be achieved upon 90% when extracting the audio digital watermark files facing the attack ways such as compression and low-pass filtering.