基于DT-CWT和SVD的鲁棒音频水印算法  

Robust audio watermarking algorithm based on DT-CWT and SVD

作　　者：张秋余[1] 许福久 赵振宇 ZHANG Qiuyu;XU Fujiu;ZHAO Zhenyu(School of Computer and Communication*Lanzhou University of Technology,Lanzhou 730050,China)

机构地区：[1]兰州理工大学计算机与通信学院,甘肃兰州730050

出　　处：《华中科技大学学报（自然科学版）》2021年第9期23-29,共7页Journal of Huazhong University of Science and Technology(Natural Science Edition)

基　　金：国家自然科学基金资助项目(61862041,61363078)。

摘　　要：在音频检索中为了利用音频水印进行信息标注,以及有效地权衡音频水印的嵌入容量、透明性和鲁棒性之间的关系,提出了一种基于双树复小波变换(DT-CWT)和奇异值分解(SVD)的鲁棒音频水印算法.首先提取原始语音的音频指纹作为音频水印;然后通过基于能量的嵌入位置选择来确定水印嵌入帧,进行4级DT-CWT后选取高频分量分段进行SVD;最后通过量化的方法在奇异值矩阵的奇异值中嵌入水印信息.实验结果表明:该算法具有较强的鲁棒性和透明性,并具有较高的嵌入容量,可以很好地满足信息标注的需要.In order to use audio watermarking to label information and to effectively balance the relationship among the embedding capacity,transparency and robustness of audio watermarking for audio retrieval,a robust audio watermarking algorithm based on dual-tree complex wavelet transform(DT-CWT)and singular value decomposition(SVD)was proposed.Firstly,the audio fingerprint of the original speech was extracted as audio watermarking.Secondly,the watermarking embedding frames were determined through the selection of the embedding position based on energy,and the high-frequency component was selected for SVD after 4-level DT-CWT.Finally,the watermarking information was embedded in the singular value of the singular value matrix by quantization.The experimental results show that the proposed algorithm satisfies the demands of labeling information well,enhances embedding capacity,and has strong robustness,transparency in comparison with the state of other methods.

关 键 词：音频水印 双树复小波变换(DT-CWT) 奇异值分解(SVD) 音频指纹 量化 

分 类 号：TP309.7[自动化与计算机技术—计算机系统结构] TN9I2.3[自动化与计算机技术—计算机科学与技术]