摘要
论文利用独立分量分析原理,分别提取隐写前后音频载体的独立分量,进行形态学变换后,以其相邻两列汉明距离的奇数阶中心矩作为特征向量,用支持向量机分类,隐写分析全局检测率可超过93%。
This paper, making use of the principle of independent components analysis, extracts independent components respectively from the original and stego audio carriers. Following morphology transformation, Odd order center moment of Hamming distance of its two neighbouring columns is used as eigenvector, and classified by support vector machine. A synthetical test accuracy of over 93% for steganalysis can be attained.
出处
《信息安全与通信保密》
2007年第6期169-170,173,共3页
Information Security and Communications Privacy
关键词
音频隐写分析
独立分量分析
形态学变换
Audio steganalysis
Independent components analysis
Morphology transform
