Deep learning is a good steganalysis tool when embedding key is reused for different images, even if there is a cover source-mismatch

摘要

Since the BOSS competition, in 2010, most steganalysis approaches use alearning methodology involving two steps: feature extraction, such as the RichModels (RM), for the image representation, and use of the Ensemble Classifier(EC) for the learning step. In 2015, Qian et al. have shown that the use of adeep learning approach that jointly learns and computes the features, is verypromising for the steganalysis. In this paper, we follow-up the study of Qianet al., and show that, due to intrinsic joint minimization, the resultsobtained from a Convolutional Neural Network (CNN) or a Fully Connected NeuralNetwork (FNN), if well parameterized, surpass the conventional use of a RM withan EC. First, numerous experiments were conducted in order to find the best "shape " of the CNN. Second, experiments were carried out in the clairvoyantscenario in order to compare the CNN and FNN to an RM with an EC. The resultsshow more than 16% reduction in the classification error with our CNN or FNN.Third, experiments were also performed in a cover-source mismatch setting. Theresults show that the CNN and FNN are naturally robust to the mismatch problem.In Addition to the experiments, we provide discussions on the internalmechanisms of a CNN, and weave links with some previously stated ideas, inorder to understand the impressive results we obtained.