Feature Selection using Mutual Information and Adaptive Particle Swarm Optimization for Image Steganalysis
Jasmanpreet Kaur, Singara Singh · 2018
The area of steganalysis has become popular for research purposes these days due to an increase in social interaction channels. The technique of revealing the hidden message embedded in the multimedia file is known as an image steganalysis. The process of an image steganalysis largely depends on the image features. Using an enormous number of features required a huge amount of time for execution and computational sources. So it is necessary to introduce a phase of pre-processing of features to improve the performance of steganalysis method. This paper focuses on feature selection process so that dimensionality of feature vector can be reduced. Feature selection technique based on Mutual Information (MI) and Adaptive inertia-weight based Particle Swarm Optimization (APSO) is proposed in this paper. First of all, differentiability of feature dimensions are calculated by using Mutual Information parameter. Then features are sorted according to the differentiability and first few features are selected. In order to improve the feature reduction phase, APSO is used further with Area Under Curve (AUC) as a fitness function. Results show that our proposed technique has shown superior results than some other similar approaches.