Compression History Analysis And Detection Of Digital Images

The authentication of digital images is attracting more and more attention from researchers. Along with the wide use of image processing softwares, it becomes more easily for people to tamper the image content which makes image forensics much harder. A lot of methods have been proposed to detect tampering process such as copy-paste, blurring, scale transformation and so on.As an important part of digital image forensics, digital image compression history detection can be used to make a first judgment of the authenticity of images. Under this circumstance, this paper focuses on the research of compression history analysis and detection of digital images. The main contribution of this thesis includes:(1) The paper gives a comprehensive introduction to the compression history detection of digital images including the researching background and existing technical literatures on the compression history detection of lossless images(BMP) and loss compressed images(JPEG). Furthermore, the paper applies the analysis and detection of image compression history to image forensics such as copy-paste detection, steganalysis and image source determination.(2) The approach of deting compression history of BMP images based on image information loss is proposed in this paper. Through analyzing the effect of JPEG compression to JPEG coefficients, statistical features reflecting image information loss difference of the detected image and the estimated image can be extracted. Thus using Fisher linear classifier and threshold division the compression history of BMP images are detected in this paper. Experiments show that the detection accuracy of the proposed approach is high, and not affected by image contents compared to Benford methods. It can also make an accurate judge of the compression history of a single BMP image without training.(3) The paper exacted features which can reflect the double compression effect including histogram features, DFT features, Benford features and Markov features of JPEG coefficients based on previous double compression detection algorithms. The original compression quality of images is usually unkown in practical use, therefore, different from exisiting literatures which use one-to-one test methods, this paper did some experiments on mixed image database with different original compress quality and comparade the results using different features.