基于音频内容的DWT域脆弱音频水印算法

Content-Based Fragile Audio Watermarking Algorithm in DWT Domain

为提高脆弱音频水印算法的安全性,实现音频内容的完整性认证,提出一种基于音频内容的可用于篡改与定位的音频水印算法.首先将音频信号分段,进行三级小波变换,然后将音频信号每段的低频小波系数的平均值映射为混沌初值,经过若干次迭代生成认证信息,并采用单个量化高频系数的方法嵌入认证水印.最后利用混沌对初值极端敏感性的特点,定位被篡改的音频位置.仿真结果表明,提出的算法对添加噪声、MP3压缩、低通滤波、重采样等常规的信号处理操作非常敏感,对恶意替换、随机剪切等去同步攻击能够报警并定位篡改位置.

The fragile audio watermarking algorithm based on the audio content is proposed to improve the security of fragile audio watermark algorithm and to realize the integrity authentication of audio content.Firstly,the audio signal is divided into segments and did the discrete wavelet transform.And then,the average value of the low-frequency wavelet coefficients of every segment is mapped into the initial value of the logistic sequence.The authentication watermark generated by iterating several times of the logistic sequence is embedded into the audio by quantifying single high-frequency coefficients.Finally,the tampered regions are located by using the extreme sensitivity to initial condition of chaos.Simulation experimental results show the algorithm has good perceptual quality and the proposed algorithm is sensitive to signal processing manipulations such as additive noise,MP3 compression,sampling rate change,low-pass filtering.Furthermore,it can detect and localize the tampered regions effectively.