A rights-based re-integration method for JPEG steganalysis
Sun Shou-jia · Chinese Journal of Liquid Crystals and Displays · 2015
Aiming at the problem of the high dimension features and complicated training in image universal steganalysis technology,this paper proposes a steganalysis method based on ensemble classifier.The method designs a ensemble classifier which combines the difference of base classifier and weighted fusion.Specifically,the algorithm uses CC-PEV as steganalysis features.Then,the method randomly structures several feature subspace,uses the bootstrap method to construct image training subsets,and train them to get a number of base classifiers.Finally,every base classifier is given weight that different from each other according to the classification result and then the final forecast is got after the combination of classification results with their respective weight.The experimental results show that using ensemble classifier for training and giving base classifiers with different weights fusion could effectively improve the accuracy of the algorithm.