基于量化频域稳健音频水印算法  

A Robust Frequency Domain Audio Watermarking Algorithm Based on Quantifying

作　　者：胡雯[1] 钟西[1] 唐向宏[1] 

机构地区：[1]杭州电子科技大学通信工程学院,浙江杭州310018

出　　处：《杭州电子科技大学学报（自然科学版）》2006年第4期9-12,共4页Journal of Hangzhou Dianzi University：Natural Sciences

基　　金：浙江省自然科学基金(M603021)

摘　　要：提出一种基于心理声学模型的频域音频水印算法。首先对音频进行MP3压缩解压,再对音频分段进行离散傅立叶变换,通过压缩前后音频功率谱的比较,然后利用心理声学模型在变换域的中低频子带域内来寻找水印嵌入位置的最佳点,水印信息的嵌入通过量化原理来完成。实验表明,该算法对一般信号处理类操作都具有较强的鲁棒性和不可听性。A high robust frequency audio digital watermarking algorithm is presented. The original signal is firstly compressed using a MP3 standard and then decompressed to the original PCM format; the both signals are transformed to frequency domain, both spectrum are computed; by comparing the two power spectrum and taking advantage of human auditory system （HAS）, we can decide the set of important marking frequencies where we can embed the watermark signal, then watermarks can be embedded into the chosen transform domain coefficient, which is based on quantization concept. The experimental results demonstrate that the proposed watermarking algorithm is greatly imperceptible and robustness against general malicious attacks.

关 键 词：音频水印 听觉模型 量化 频域 

分 类 号：TP391.41[自动化与计算机技术—计算机应用技术]