Investigation of Steganalysis Techniques
Steganography, the concealment of data within other carrier media, represents a typical security issue to the protection of sensitive information as well as the collection of intelligence. On the contrary, Steganalysis aims to identify the presence of suspected objects in media and if possible, destroy or recover the hidden information. Within the repeating no "end all" competition between these counter-techniques, the latter usually takes a "passive" role due to its inherent technical difficulties. In this thesis, we carry out a thorough review of the fundamental implications associated with steganography such as the history of steganography, steganographic framework, and the embedding process. In practice, many different media have been chosen as potential carriers (text, video, and digital images for example) but digital images are still the most popular carrier medium. Least-significant bit (LSB) insertion, which provides a high embedding capacity, is the most widely-implemented and easiest approach. We introduce various technical challenges that steganalysis has and alternations available in steganalysis in order to achieve its goal. We find the detection of the hidden message is the most desirable outcome for this technique. We classify the steganalytic domain into two categories: Algorithm Specific steganalysis and Universal steganalysis. For each category, we investigate some typical algorithms. We analyze the theoretical bases that make them workable. Apart from this, we discuss the strengths and weaknesses associated with each algorithm as well as improvements that have been made in subsequent research. In contrast to the in-depth investigation of either steganography or steganalysis techniques, this thesis also brings these theories into practice by way of performing experiments towards two representative steganalytic methods: Raw Quick Pair analysis and RS analysis. Through the critical reviews and experiments, we aim to validate them through experiments and highlight points that are not addressed in their initial proposals.