Research On Text Steganography And Steganalysis

Information hiding has become a new hot interest area in the research of information security in the network age and has wide potential applications in many fields such as copyright protection of digital media, hidden communication. Steganography and watermarking are two main sub-disciplines of information hiding. Watermarking is mainly used for copyright protection, and has been deep researched and applied widely. Steganography is developed for hidden communication which hides the existence of communication.Steganography can be used to transfer some baleful messages when it is used to protect our information. Thus it is necessary to filter or intercept that baleful information, which is the role of steganalysis. And steganalysis is also useful for steganography, which can help us to design more secure steganographic algorithms.This thesis mainly focuses on the art of embedding messages and also the techniques of detecting hidden messages in texts. It closely keeps up with international research progress and the demands in practice. Based on the anatomy of structure and redundancy of text data which shows the feasibility of embedding messages in texts, several improved or new practical steganographic algorithms are presented after analyzing the shortages of the existing ones. Before probing into detection techniques, the thesis discusses the network data filtering techniques and proposes a hardware multi-pattern matching algorithm based on SBF. Then we present several effective steganalytic algorithms based on in-depth anatomy of current steganographic software and algorithms. Furthermore, the thesis explores extracting messages from the stego-texts, and investigates some extracting techniques against those simple steganographic algorithms. The experiments and applications prove the efficiency and values of our research.The main contributions of this thesis are summarized as follows:1. A new steganographic technique used in hypertext is proposed. Based on the analysis of HTML protocol, we find that HTML browser is sensitiveless to the upper or lower letters of markup. We proposed a method that embeds messages into HTML by changing the states of letters according to that.2. A steganographic algorithm of hiding messages in IP addresses of IP packets is presented. The existing methods embedding messages in IP packets is not secure enough since they embed information in those redundant or optional fields, and can easily be attacked. Our algorithm is more secure by embedding messages in the IP addresses since they should not be changed and filterd. 3. A new hardware multi-pattern matching algorithm based on SBF is proposed. Experiments and analysis show that our algorithm is effective for both large group of patterns and high speed.4. Several detecting techniques are proposed based on the anatomy of inherent features of existing steganographic software and algorithms.5. Five statistic detecting techniques are proposed against semantic steganographic methods based on statistical analysis by modeling the natural text. Experiments show high efficiency and reliability of the techniques.6. A blind detecting technique based on SVM is proposed. In this algorithm, the SVM classifier composed of a vector combined with several statistical features is trained with sample data and then is used to classify the texts. Experiments show that the SVM classifier is more effective than those techniques only using single feature.7. Explores extracting messages from stego-texts, and presents a technique of estimating the capacity of embedded messages for dictionary-steganography. Experiments show its high precision.The thesis discusses not only in theory, but also in practice, and proves the efficiency and reliability of our ideas and algorithms with enough experiments.Finally, we summarize our research work, and analyze the problems and discuss further research topics and directions in the future.