Image steganalysis based on multi-feature extraction using support vector machines
Renbin Zhang · Journal of Hefei University of Technology · 2007
The statistical features of high frequency wavelet sub-band coefficients of natural images and stego-images are analyzed,and the feature vectors are extracted from the texture statistical moment,the DCT coefficients' histogram moment and correlativity between blocks,and then the stego-images are classified by the SVM method.The proposed steganalysis algorithm extracts multiple features from different types and perspectives,thus solving the insufficiency of feature extraction in current universal steganalysis methods.Experiment results show that the new algorithm is an effective universal detection method with high accuracy.