After Processing Operation For The Blind Type Of Steganalysis Trustworthiness Evaluation

Digital image steganography hides data into redundancy fields of image, the development of steganography brings about the progress of steganalysis.The diversity of steganalysis makes it more and more important to evaluate the performance of steganalysis. On one hand, evaluation on the performance of steganalysis analyze strong points of steganalysis; on the other hand, it points out shortcomings of steganalysis.Evaluation on the performance of steganalysis provides us ways to choose from steganalysis and improve steganalysis.In this dissertation, we evaluate on the performance of steganalysis based on processed cover and stego images, and propose the conception of blind steganalysis credibility which is based on the type of stego images. The contributions of this thesis are listed as follows:(1) A brief summary is made on concept, model and classification and development of steganalysis and its evaluation methods. Reasearch and analyzes the principle and characteristics of several typical staganography and staganalysis methods;(2) An introduction is made on image processing methods with normal photoed images. An experiment is made to obtain false alarm rate and detection rate of steganalysis before and after image processing.Based on experiments, we analyze the influence of image processing on steganalysis;(3) Based on DS Evidential Theory, combined with experiment results, we give detailed ways to calculate steganalysis credibility and resolve conficts of evidence. Steganalysis credibility quantifies the influence of image processing on staganalysis, we explain the causes of the influence on the basis of credibility.After that, we point out that we have better calculate steganalysis credibility after digital forensics which can recongnize different image type.Finnally, a conclusion is drawn on the shortcomings of this dissertation, and the contents which is worthy of further research on the topic are also prospected.