摘要
论文利用主元分析原理,分别从时域、小波域、频域几方面来提取音频载体隐写前后的不重要主元,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,隐写分析分类准确率可超过95%。
This paper makes use of the principle of principal component analysis, and extracts least significant principal components of original and stego audio from time and wavelet and spectrogram domain respectively. Following morphology transformation, Odd order center moment of Hamming distance of its two neighbouring columns is used as eigenvector, and classed by support vector machine. The classification accuracy of steganalysis can be attained over 95%.
出处
《信息安全与通信保密》
2007年第2期63-64,67,共3页
Information Security and Communications Privacy
关键词
音频隐写分析
主元分析
形态学变换
Audio steganalysis
Principal component analysis
Morphology transform
