基于广义高斯模型的音频篡改盲检测  

Blind audio tamper detection based on generalized Gaussian distribution

作　　者：林晓丹[1] 

机构地区：[1]华侨大学信息科学与工程学院,福建厦门361021

出　　处：《计算机工程与设计》2013年第9期3050-3053,3062,共5页Computer Engineering and Design

基　　金：泉州市科技计划基金项目(2011G7)

摘　　要：针对音频篡改被动取证的问题,提出了一种基于改进离散余弦变换(MDCT)域的盲检测方法。音频信号MDCT系数直方图可用广义高斯分布模型进行拟合。原始音频的广义高斯模型形状参数集中分布在一定的区间内,而篡改后的音频信号在篡改发生边界,其广义高斯模型的形状参数将发生明显变化,通过设置形状参数阈值区分音频信号是否遭受篡改。实验结果表明,对于改变音频内容的删除、替换和拼接操作,该方法在篡改检测和定位方面具备有效性。A blind detection method is proposed for audio tampering via modified discrete cosine transform （MDCT） domain anal- ysis. The distribution of MI^T coefficients for untampered audios can be well fitted by generalized Gaussian distribution with shape parameters in a certain level. Whereas a dramatic change will occur in the shape parameters as tampered audio is con- cerned. Thresholds are set to determine whether an audio signal is tampered or not. Experimental results show the efficacy of the proposed method in detecting and locating audio tampering under forgeries like deletion, substitution and splicing.

关 键 词：篡改检测 被动取证 改进离散余弦变换 广义高斯分布 形状参数 

分 类 号：TP309[自动化与计算机技术—计算机系统结构]