| With the fast development of computer and network technology, digital multimedia data is being transmitted more and more conveniently and frequently on communication network. When bringing conveniency to communication, it also brings new challenge to communication security. Information hiding has become the hotspot in reseach of network security. Steganography, as the main embranchment of information hiding, hiding the secret information into open digital media such as images, audios, videos for the purpose of secret communication. With steganography utilizing in the military and commerce ,it is also used by criminals which poses a threat to national security. The purpose of digital image steganalysis system proposed in this paper is to detect and discriminate information hiding in digital images, so it has important practical significance.This thesis reviews the information hiding and image steganography techniques, describes several current and popular steganographic methods and analyzes the advantages and disadvantages of each method. A steganalysis method for JPEG images based on DCT domain statistic features and DCT coefficient difference is proposed in this paper, and experiment has validated the effectiveness of this method in different embedding rates., especially at higher embedding rates it can reach high detection accuracy.In this steganalysis system, by researching the statistical properties of the image DCT coefficients, Laplacian statistical model of the DCT coefficients for JPEG images is established, and the parameters was effectively estimated; in feature extraction, by researching DCT Coefficient of difference histogram of images, we not only retain the mean, variance, kurtosis, skewness and other characteristics, but also introduce the difference of DCT coefficients, and through several experiments demonstrate the effectiveness of the method; in the choice of classifier , support vector machine classifier is selected and we apply it to the steganalysis system, for the steganalysis detection (whether the steganography is present in real terms is the two types of pattern recognition of information classification problem).This paper designs the JPEG image steganalysis platform, which is based on VC + + programming language, can achieve a variety of batch steganography detection of JPEG image steganalysis method. |
