一种双重判断机制的音频篡改盲检测算法  

A Blind Audio Forgery Detection Algorithm Based on Dual Judgment Mechanisms

作　　者：吕志胜[1] 胡永健[1] 李晗[1] 刘琲贝[1] 

机构地区：[1]华南理工大学电子与信息学院,广东广州510640

出　　处：《华南理工大学学报（自然科学版）》2014年第8期70-76,共7页Journal of South China University of Technology(Natural Science Edition)

基　　金：国家"973"计划项目(2011CB707003);广东省自然科学基金团队项目(9351064101000003);华南理工大学中央高校基本科研业务费专项资金资助项目(2014ZZ0036)

摘　　要：插入和删除是两种常见的音频篡改操作.针对现有基于电网频率(ENF)信号的音频篡改检测算法对插入和删除操作定位精度不高的问题,提出一种双重判断机制的篡改盲检测算法.机制一利用最大相关偏移量曲线来确定篡改位置;机制二根据最大相关偏移量曲线所对应的斜率曲线来确定篡改位置.将机制一和机制二联合使用可获得更好的篡改定位精度.为了简化算法实现,文中还提出一种不用引入额外ENF参考信号计算最大相关偏移量的方法.与现有文献中两种代表性算法相比,文中算法具有更高的篡改定位准确度,且对重采样、压缩及加噪这3种常见的音频处理操作有一定的鲁棒性.Insertion and deletion are two commonly-used audio forgery operations,and the current Electric Network Frequency（ ENF）-based audio forgery detection algorithms cannot accurately determine the location of insertion and deletion operations. In order to solve this problem,a blind forgery detection algorithm based on dual judgment mechanisms is proposed in this paper. In the investigation,the first mechanism utilizes the information from a MOCC（ max offset for cross correlation） curve to locate the forgery while the second mechanism estimates forgery regions according to the slope curve which corresponds to the MOCC curve. The combined use of the two mechanisms can obtain a higher forgery localization precision. For simplicity of implementation,a method is put forward to calculate the MOCC without an extra reference ENF signal. As compared with two current representative methods,the proposed detection method can locate the deletion and insertion operations more accurately. Besides,it has reasonable robustness when applied to such common audio operations as re-sampling,compression and noise addition.

关 键 词：音频篡改检测 电网频率 最大相关偏移 斜率 双重判断机制 

分 类 号：TN911.23[电子电信—通信与信息系统]