Research On Text Steganography Based On Low Distortion Substitution

With the increasing popularization of Internet, people's communication gradually relies on the network. There are many information transmissed through the network by the form of digital every day. The problem of information and network security is more and more serious, that makes the technology of information hiding rapidly developed. Text steganography is one kind of the information hiding technology, it use text as the carrier of convert communication to carry secret information by hiding its existence. Most existing text steganographic methods have the problem of poor security. Therefore, in this thesis, taking the synonyms in the text as the steganographic carrier, and the distortion caused by synonym substitution as the measures or performance, more practical, secure, low distortion steganographic schemes are proposed. The main results in this thesis are as follows:(1) As the substitutions of synonyms in different locations would cause different degree of distortion, this paper proposes a text steganographic algorithm based on prior substitution with low distortion, which can effectively reduce the overall distortion. The algorithm firstly selected just two words which have the highest fitness compared with other words from each synonymous word group, and took them form a substitution combination with low distortion. During the process of steganography and in the global view, all synonyms substitution combinations in carrier text will be sorted by the distortion degree caused by synonym substitutions, and then in the sort order synonym substitution operations with the small distortion degree of combination are be prior chose to embed information, to reduce the change of statistical properties of the synonyms before and after embedding. The experimental results demonstrate that the algorithm can effectively reduce the influence raised by substitutions in the text as a whole.(2) Because of the low resistance ability of text stegoanalysis and the low embedding efficiency of the most existing text steganography, this paper proposes a new steganographic algorithm with minimizing substitution distortion. The algorithm analyzes the frequency characteristics of synonyms and introduces word frequency tomeasure the distortion for designing the distortion function firstly, and then makes use of Syndrome-Trellis Codes(STC) to select the best modification path, so that the distortion introduced into the text can tend to minimize in theory. Thus, the number of modification can be keep in a small range to attain low distortion. The experimental results demonstrate that the algorithm can achieve minimum distortion when the embedding rate is appropriate.