Mismatched Steganalysis Based On Image Source Identification

In the information age,with the rapid development of communication technology,information security issues such as privacy protection gradually get people's attention,information hiding is to embed the secret information into digital media files to transmission to achieve the purpose of covert communication while steganalysis is to detect whether digital media contains confidential information.The opening of information hiding method allows criminals to easily use these technologies to pose a threat to social and even national information security.So the research of steganalysis that can detect whether there is information hiding in the huge amounts of data in network is very important.JPEG image storage format is one of the most widely used image formats.Traditional JPEG image steganalysis is based on the hypothesis that the training sample and test sample has the similar statistical properties and characteristics of distribution.But in reality,the hypothesis does not stand,which can lead to JPEG image mismatch,JPEG image steganalysis performance reduction.There are a lot of literature points out that the seriousness of the mismatch problem while there are few concrete methods to solve this problem.This paper studies the JPEG mismatch steganalysis from the the angle of the JPEG source image mismatch problem.(1)Firstly,the background,research significance and overseas and domestic research status of JPEG image steganalysis are introduced,then the traditional JPEG steganalysis framework is introduced,and then this paper analyses the reasons and effects of the JPEG mismatch steganalysis,and this paper gives a simple summary of the existing methods such as transfer learning and multitasking learning of solving mismatch problems and proposes a new framework based on source identification.(2)This paper simply introduces the JPEG compression,decompression and double compression and proposed a new JPEG double compression detection model based on fusion features,and this paper has carried on the contrast experiments of fusion features,Benford features and GLCM features and confirm the effectiveness of this algorithm.(3)This paper proposed a new source camera identification model based on JPEG double compression detection to reduce the effects of JPEG image double compression on source identification.And the accuracy is improved.(4)This paper proposed a JPEG image mismatched steganalysis model by combining JPEG image steganalysis with source identification.Firstly this paper detect the test imagewhether it is JPEG double compressed,and carries on source identification on the test image to find its source,then this paper train JPEG steganalysis model with the train image which is from the same source camera with the test image,and finally find out whether the test image hiding information which solve the problem of JPEG image steganalysis source mismatch.