摘要
提出了一种基于小波包分解和多类支持向量机分类的音频隐秘检测算法,该算法首先对音频文件进行小波包分解,然后根据小波分解系数绝对值和绝对值线性预测的误差生成特征向量,并采用多类支持向量机进行模式分类。在不同嵌入率下对几种常见的隐秘软件生成的隐秘音频进行仿真试验,结果表明,该算法具有较强的通用性,对于隐密音频文件具有较高的识别率。
A new detection of steganography algorithm based on wavelet statistics of audio and multi-class SVM is proposed.In this algorithm,wavelet packet decompision is implemented to audio files,the magnitude of decomposition coefficients and the error between the actual coefficient and the predicted coefficient magnitudes are used to yield statistics.The multi-class support vector machine algorithm has been employed in the pattern discrimination.Stego audios created by several tools are tested under different embed rates.The experimental results show that the algorithm has stronger universal performance and higher discriminating rate.
出处
《计算机工程与应用》
CSCD
北大核心
2007年第34期162-164,共3页
Computer Engineering and Applications
基金
国家自然科学基金(the National Natural Science Foundation of China under Grant No.60473029)
信息安全教育部重点实验室课题资 助项目(No.200409)
武警部队军事科研项目。
关键词
多类支持向量机
小波包分解
隐密音频检测
multi-class SVM
wavelet packet decomposition
dection of steganography audio
