Image steganalysis based on co-ocurrence matrix

A new steganalysis scheme for gray images was proposed in this paper. A total of 180 dimensional feature vectors were derived from the cooccurrence matrix,which was calculated directly from horizon and vertical differential image. SVM was used to classify the 180 dimensional feature vectors. The experimental results for 3 kinds of popular steganographic schemes (Cox spread spectrum,QIM,LSB)have demonstrated that the proposed scheme outperforms the existing steganalysis techniques.