High Performance Image Steganalysis Through Stego Sensitive Feature Selection Using MBEGA

Steganalysis has emerged as an important branch in information forensics. Due to the large volumes of security audit data as well as complex and dynamic properties of steganogram behaviors, optimizing the performance of steganalysers becomes an important open problem. This paper is aimed at increasing the performance of the steganalysers in through feature selection thereby reducing the computational complexity and increase the classification accuracy of the selected feature subsets. In this study, we propose to employ Markov blanket-embedded genetic algorithm (MBEGA) for stego sensitive feature selection process. In particular, the embedded Markov blanket based memetic operators add or delete features (or genes) from a genetic algorithm (GA) solution so as to quickly improve the solution and fine-tune the search. Empirical results on suggest that MBEGA is effective and efficient in eliminating irrelevant and redundant features based on both Markov blanket and predictive power in classifier model. Experimental results prove that the proposed method is superior in terms of number of selected features, classification accuracy, and running time than the existing algorithms.