Audio Steganalysis Based on Principal Component Statistics
Li Guo · China Information Security · 2007
This paper makes use of the principle of principal component analysis, and extracts least significant princi-pal 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%.