JPEG Images Steganalysis Based on Statistic Characteristic and SVM
Jing Sun · Jisuanji gongcheng · 2008
This paper presents a steganalysis scheme which can effectively attack JPEG steganographic.The method exploits the first and the second order statistic characteristics.It extracts these characteristics of histogram and the partial statistical characteristics of intra-block and inter-block from the cover-images and the stego-images respectively.In order to achieve the purpose of general detection,it uses Support Vector Machine(SVM)to train and distinguish these characteristics,and establishs the model.Experimental results show that the method performs well in steganalysis methods such as Jsteg,F5,OutGuess,and MB.