上海大学学报(自然科学版) ›› 2016, Vol. 22 ›› Issue (4): 388-397.doi: 10.3969/j.issn.1007-2861.2014.05.017

• 通信与信息工程 • 上一篇    下一篇

基于EVD变换的鲁棒音频水印算法

童人婷1, 程航1,2, 张新鹏1   

  1. 1. 上海大学通信与信息工程学院, 上海200444;
    2. 福州大学数学与计算机科学学院, 福州350108
  • 收稿日期:2014-11-26 出版日期:2016-08-30 发布日期:2016-08-30
  • 通讯作者: 张新鹏(1975—), 男, 教授, 博士生导师, 博士, 研究方向为多媒体信息安全. E-mail: xzhang@shu.edu.cn
  • 作者简介:张新鹏(1975—), 男, 教授, 博士生导师, 博士, 研究方向为多媒体信息安全. E-mail: xzhang@shu.edu.cn
  • 基金资助:

    国家自然科学基金资助项目(61472235)

Robust audio watermarking based on eigen-value decomposition

TONG Renting1, CHENG Hang1,2, ZHANG Xinpeng1   

  1. 1. School of Communication and Information Engineering, Shanghai University, Shanghai 200444, China;
    2. College of Mathematics and Computer Science, Fuzhou University, Fuzhou 350108, China
  • Received:2014-11-26 Online:2016-08-30 Published:2016-08-30

摘要:

常见的数字信号处理往往会改变音频信号的高频分量并引入随机噪声,并且易造成数字水印信息的位置改变. 提出了一种新的数字音频水印算法. 在该算法中, 原始音频被分为两部分: ① 运用量化索引调制来嵌入伪随机序列生成的二值同步码; ② 利用特征值分解(eigenvalue decomposition, EVD)方法先对离散小波变换(discrete wavelet transform, DWT)低频系数进行变换, 然后在生成的对角阵中用量化索引调制嵌入水印信息. 实验结果表明, 在确保不可感知性和较强鲁棒性的前提下, 可大幅度提高水印嵌入容量, 达到172 bit/s.

关键词:  高容量,  鲁棒 ,  特征值分解 , 音频水印

Abstract:

Common digital signal processing often introduces noise into audio signals and cause high-frequency distort. Meanwhile, both signal processing operations and malicious attacks may change location of watermark information. By making use of robustness of eigen-value decomposition (EVD), a blind audio watermarking algorithm is proposed. The original audio signal is divided into two parts. Binary codes for synchronization are embedded into the first part using quantization index modulation (QIM). The approximation components of discrete wavelet transform (DWT) of the second part is transformed using EVD to generate a diagonal matrix, and the watermark information is embedded into the matrix entries with QIM. Experimental results show that embedding capacity of the proposed method is as high as 172 bit/s, and it still maintains good audio quality and can tolerate a wide range of common attacks.

Key words: audio watermarking, eigen-value decomposition (EVD), high capacity, robust