Research on Holism-Based Feature Extraction and Fusion for Steganalysis
Bo Wang · Dianzi xuebao · 2006
The extraction and fusion of statistical features has been a key problem demanding solving for steganalytic technology in the context of the prisoner's problem presented by Simmons and Cachin's security theory model.We divide the statistical features into two categories,i.e.,inherited features and introduced features,and the importance of these features is further discussed.Furthermore,based on the research of steganalytic methodology and the theory of holism,a fusion strategy of statistical features is proposed,and the availability of the proposed fusion strategy is verified by our image steganalysis system,which is developed using the strategy.