Multi-class JPEG Steganalysis using Extreme Learning Machine

This paper proposes a novel Multiclass Steganalysis process, for colored JPEG images using Extreme Learning Machine (ELM) as classifier. The feature set used for classification of images consists of 810 features consisting of 405 Markov features and 405 calibrated Markov features. The Markov features are based on Markov random process applied on correlations among JPEG coefficients of the image. The calibrated Markov features are the difference between the Markov features of the image and Markov features of a reference image, obtained by decompressing, cropping and recompressing the image. It is evident from the experiments that our proposed multi-class steganalysis method show good performance results in classifying the stego-images into their respective classes; these classes correspond to the embedding steganography techniques. The other advantage of the proposed method is its fast speed which is due to ELM as the learning process of ELM is very fast.