In this paper, an unsupervised steganalysis method that combines artificial training sets and supervised classification is proposed. We provide a formal framework for unsupervised classification of stego and cover images in the typical situation of targeted steganalysis (i.e., for a known algorithm and approximate embedding bit rate). We also present a complete set of experiments using 1) eight different image databases, 2) image features based on Rich Models, and 3) three different embedding algorithms: Least Significant Bit (LSB) matching, Highly undetectable steganography (HUGO) and Wavelet Obtained Weights (WOW). We show that the experimental results outperform previous methods based on Rich Models in the majority of the tested cases. At the same time, the proposed approach bypasses the problem of Cover Source Mismatch -when the embedding algorithm and bit rate are known-, since it removes the need of a training database when we have a large enough testing set. Furthermore, we provide a generic proof of the proposed framework in the machine learning context. Hence, the results of this paper could be extended to other classification problems similar to steganalysis.

This paper presents a novel method for detection of LSB matching steganography in grayscale images. This method is based on the analysis of the differences between neighboring pixels before and after random data embedding. In natural images, there is a strong correlation between adjacent pixels. This correlation is disturbed by LSB matching generating new types of correlations. The presented method generates patterns from these correlations and analyzes their variation when random data are hidden. The experiments performed for two different image databases show that the method yields better classification accuracy compared to prior art for both LSB matching and HUGO steganography. In addition, although the method is designed for the spatial domain, some experiments show its applicability also for detecting JPEG steganography.

In this paper, a methodology to detect inconsistencies in classi cation-based image steganalysis is presented. e proposed approach uses two classi ers: the usual one, trained with a set formed by cover and stego images, and a second classi er trained with the set obtained a er embedding additional random messages into the original training set. When the decisions of these two classi ers are not consistent, we know that the prediction is not reliable. e number of inconsistencies in the predictions of a testing set may indicate that the classi er is not performing correctly in the testing scenario. is occurs, for example, in case of cover source mismatch, or when we are trying to detect a steganographic method that the classi er is no capable of modelling accurately. We also show how the number of inconsistencies can be used to predict the reliability of the classi er (classi cation errors).

No abstract