A Blind Steganalysis Method Based on PSOSVM
Jianjun Wang · Information and Electronic Engineering · 2009
To study the feature selection in blind steganalysis, a new feature selection method based on Particle Swarm Optimization and Support Vector Machine(PSOSVM) is proposed. Using nonlinear SVM as classifier, this method employs the Particle Swarm Optimization(PSO) algorithm to find the best image feature sets as training and testing sets and chooses the best Support Vector Machine(SVM) parameters at the same time. Then the selected image feature sets and parameters are used to detect the stego-images. In order to demonstrate its validity,the proposed method is compared with several existing methods by experiment. The experimental results show that the proposed method outperforms the Farid, Analysis of Variation(ANOVA) and F-score methods. It has higher recognition ratio of stego-images and improves the detection efficiency.