用于版权管理的数字音频水印算法  

Digital Audio Watermarking Algorithm for Media Copyright Management

作　　者：赵学敏[1] 郭宇弘[1] 邹学强[1] 刘建[1] 颜永红[1] 

机构地区：[1]中国科学院语言声学与内容理解重点实验室,北京100190

出　　处：《电子与信息学报》2011年第10期2384-2389,共6页Journal of Electronics & Information Technology

基　　金：国家自然科学基金(10925419;90920302;10874203;60875014;61072124;11074275)资助课题

摘　　要：该文提出了一种用于多媒体资料版权管理的自同步音频盲水印算法。由于版权应用需要通过解码结果得到素材的使用情况,对解码结果时间精度有一定要求,同时对鲁棒性和不可感知性要求也较高。针对之前算法存在鲁棒性不够,时间精度低的问题。该文提出的算法利用扩频的方法和音频信号的短时相关的特性提高嵌入水印的鲁棒性的同时,通过心理声学模型的应用改善算法的不可感知性。算法通过合理设计了水印信息帧结构和同步策略提高了解码结果的时间精度。实验结果表明,该文的方法在具有很高不可感知性的情况下,能够有效地抵抗各种常见的音频编辑操作。A blind audio watermarking algorithm with self-synchronization for copyright protection is proposed. In the application of media copyright management, the decoding results should have a time precision, better robustness and fidelity. However, most of algorithms can not provide enough robustness and time precision. To deal with these problems, in this paper, spread spectrum theory and the short-time relevance of audio signal are used to generate a watermark. To improve the fidelity of audio, a psychoacoustic auditory model is further used. Moreover, special synchronization strategy and watermark message frame structure are used to get a high time precision decoding results. Experimental results show that the watermarking scheme is robust against common audio editing attacks while keeping the watermarked audio a high perceptual quality.

关 键 词：数字水印 心理声学 扩频 版权保护 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]