摘要
提出了一种MDCT(修正的离散余弦变换)域的MP3隐写分析方法。该方法通过改进的Markov特征集和联合密度特征集提取出MDCT系数的特征,经过方差分析选择有效的特征用于支持向量机(SVM)的训练与检测。实验结果表明,与以往方法相比提高了对MP3Stego的检测率,能有效地检测常见的MDCT系数修改的MP3信息隐藏方法。
A steganalysis method in MP3 based on MDCT features is proposed. The original Markov and joint den- sity features are expanded to the inter-frame and intra-frame of MDCT coefficients. By selecting the effective fea- tures by ANOVA, Support Vector Machine(SVM)is used to distinguish between cover MP3 and stego-MP3. Exper-imental results show the reasonable improvement of detection performance of MP3Stego and successful detection of the existence of the secret data in MP3 audio for MDCT modification.
出处
《计算机工程与应用》
CSCD
2012年第10期115-119,共5页
Computer Engineering and Applications