A Theory Of Security In Steganography

Steganography is a kind of a long history of secret communication technology, which protects the security of confidential information by hiding its existence. With the development and popularity of Internet technology, the carrier of secret has turned into digital multimedia information from material object. This change gives new meaning to Steganography. As a secret communication technology, the security research in this area is undoubtedly the core issue. This dissertation proposes the source of the security in steganography using axiomatic method. Combining steganography with source coding, a systematic study on has been made on the aspects of the theory of security for steganography, the measure of security for steganography and the design of steganographic method. The main contribution of this dissertation is as follows.About the theory of security of steganography:First, the representation of information doesn't have specific meaning, but a fixed representation of information can mean anything with suitable decoder. Secondly, statistically, the total entropy of changeable variable must be equal to the information quantity of secret. At last, the existence of secret only can be found by denying the proposition that the sample didn't come from cover source. And because of the objective randomness of cover source, the undetectable change is ubiquity in cover information. Hence, the security in steganography derived from the objective randomness of cover source, and a new theory of security for steganograpy based on value redundancy is proposed.About the measure of security of steganography:Beccause the only way to find the existence of secret is denying the proposition that the sample didn't come from cover source. Based on the generalized method of types, using the upper limit of minimum probability of n-order statistic defines the security of stego-information, in the test of n-order statistic for n-order statistical constraints. Due to the probability density function is the only constraint of source, the divergence of probability density function between stego-source and cover-source can measure the security of steganographic method. Then according to the necessity of the measure of steganographic method, the exclusive form of the divergence of probability density function for steganographic method is presented by functional analysis. Using the above conclusions, the limitations in security due to the natural cover source statistical rules approach with inductive method are proposed, and the quantitative relationship among embedding rate, sample size and the security is also proposed. About the steganographic method:According to the new knowledge about security in steganography, the root cause of type and relevance change in stego-information is the lack of statistical rules of natueal cover source. Using new internet media as cover-infromation, two security steganographic methods are presented.