变换域内利用SVD的音频数字盲水印算法

A Digital Audio Blind Watermarking Algorithm Using SVD in Transform Domain

为了有效地保护音频数字的版权,提出了一种基于离散小波变换(Discrete Wavelet Transformation,DWT)和奇异值分解(Singular Value Decomposition,SVD)并利用人耳听觉系统(Human Audio System,HAS)特性的音频数字水印算法。根据音频信号的特征选择短时能量大的音频段,利用局部最大值追踪算法选择最佳嵌入位置,并进行DWT变换,提取中频系数做奇异值分解,采用量化的思想嵌入水印。在提取水印阶段,只需要判断奇异值的奇偶性即可得到水印序列。验证结果表明,该算法具有良好的透明性和鲁棒性,对于常见的攻击都有较强的抵御能力。

For copyright protection of digital audio,a Discrete Wavelet Transform(DWT)-Singular Value Decomposition(SVD)based blind audio watermarking algorithm exploiting the Human Audio System(HAS)features is proposed.The algorithm finds the audio segments with highest short-term energy according to the characteristics of the audio signal.The optimal embedding position is selected from those segments by a local maximum tracking algorithm and the DWT is applied.The intermediate frequency coefficients are extracted for SVD,and the watermark is embedded based on quantization.In the extraction process,the watermark sequence can be obtained by judging the parity of the singular values.Validation results show that the proposed algorithm is transparent,robust and resistance to common attacks.