基于均值量化的多目的数字音频水印算法

Mean-quantization-based multipurpose digital audio watermarking scheme

为了达到版权保护和内容认证的双重目的,提出一种新颖的多目的水印嵌入算法.首先介绍了均值量化(mean-quantization)方法嵌入水印比特信息的基本原理,在此基础上提出了数字音频信号的多目的水印嵌入算法,该算法在对数字音频信号作离散小波变换基础上,充分利用均值量化的优势,将视觉可辨的水印—二值图像嵌入小波低、中频系数中,实现了鲁棒性水印和脆弱性水印同时嵌入音频信号中,达到版权保护和内容认证的双重目的.实验结果表明,文中提出的多目的水印算法具有很强的鲁棒性和较好的脆弱性.

A novel multipurpose digital audio watermarking scheme is proposed in which robust and fragile watermarking are simultaneously embedded for copyright protection and content authentication. By mean-quantization, two complementary watermarks were embedded in a digital audio signal＇s wavelet coefficients. In order to enhance robustness, robust watermarks were embedded in the lowest frequency components of a digital audio signal, while fragile watermarks, which are used to determine whether the digital audio signal has been altered, were embedded at the intermediate frequency. Experimental results show that the performance of the multipurpose watermarking scheme is indeed superb in terms of robustness anti fragility.