JPEG Steganalysis Of YASS

As a novel and effective solution for covert communication and copyright protection, information hiding aims to conceal the content of the communication and make the message unintelligible to those unauthorized. It claims higher security compared with cryptography as a solution for covert communication. If the technology is applied by terrorists, it will endanger the societies' secure and harm the country's benefit. Therefore it's important to research on detecting the hiding information which is steganalysis, and the research on steganalysis can help to gain the better steganography arithmetic.Since JPEG is pervasive used on the Internet and digital image processing area, JPEG images are an ideal cover media for information hiding. YASS is a new secure JPEG steganographic algorithm having the ability of undetection using current best blind steganalysis. This thesis mainly studies the JPEG steganalysis technology of YASS.Focuses on the characteristic of YASS' random embedding, this paper studies the locaton of embedding blocks using the characteristic of YASS' random embedding, and finds out that the YASS' embedding processing will have an impact on JPEG images' statistical property. Based on the discussing of the statistical property, an effective special steganalytic algorithm which use several features and the embedding datas' probability distribute is proposed.Afte studing some limications of the special algorithm, this paper analyzes the effects on correlation statistical property of JPEG image's pixels and DCT due to YASS ,based on which, the paper propose a steganalytic algorithm with Markov features, correlative variance features, co-occurrence features, and using SVM classifier. Experiments show that the algorithm can effectively detect the images which have been produced by all of YASS, and have a lot of applied value.At the end of this thesis, the prospect of YASS' steganalysis is discussed as well as the potential ways for further research.