Abstract
A promising steganograhic method-Yet Another Steganographic Scheme(YASS) was designed to resist calibration based blind steganalysis via embedding data in randomized locations. The existing steganalysis methods analyze it ineffectively or use high-dimensional feature set or are targeted steganalysis methods. In this paper, we present a steganalysis method of lower-dimensional feature sets, and it can effectively detect YASS. The 198-dimensional feature vector is calculated in the wavelet domain as statistical moments of wavelet characteristic function and Markov process features of low frequency coefficients. A SVM based classifier is trained on the extracted features for the detection of the presence of steganography. Experimental results show that the new feature set provides significantly better results for detecting YASS than previous art.