摘要
提出了一种复倒谱域数字音频盲水印算法。该算法将音频信号分为包含相同采样点的若干段,计算指定段信号的均值,对均值小于0的段的信号取反,对均值大于等于0的段保持不变,然后对所有指定段进行复倒谱变换,将这些段的复倒谱变换系数均值与设定的阈值作比较,结合水印序列的值,通过增加、减少或不改变复倒谱系数的均值来嵌入水印。水印提取为盲提取。实验结果证明该算法具有较强的鲁棒性与不可知觉性,可以抵抗常见信号处理的攻击。
A novel digital audio blind watermark algorithm based on complex cepstrum domain is presented. In the algorithm, the audio signal is divided into several sections which include same samples, and the mean value of the signals of each appointed section is calculated. The signals of the sections whose mean values are less then 0 are reversed. Then complex cepstrum transform are implemented for all appointed sections. According to the comparison between the mean values of the complex cepstrum coefficient of these sections and the pre-determined threshold, and com- bining with the values of watermark sequences, watermark is embedded by increasing, reducing, or keeping the mean value of the complex cepstrum coefficient. The watermark can be extracted without original audio signal. The experimental results show that this algorithm is robust and imperceptible and can withstand attacks of common signal processing.
出处
《北京石油化工学院学报》
2008年第1期1-5,共5页
Journal of Beijing Institute of Petrochemical Technology
关键词
数字音频水印
复倒谱变换
鲁棒性
不可知觉性
digital audio watermark
complex cepstrum transform
robust
imperceptibleness
