The Research Of Steganalysis For 2LSB Replacement Steganography

Information hiding is an important technique to make safe transmission of information. However, the hostile forces and the terrorists can also use it to undermine social stability and endanger national security. Therefore, how to detect hidden information in order to prevent some people from communicating with open channel has practical value. This thesis studies the steganalysis questions for 2LSB (Two Least-Significant Bits) replacment embedding in space domain of image. The main contents are as follows:1) A steganalysis method which referred as ERS method for 2LSB is proposed. Based on the principle of finite state machine, the relationship between the embedding ratio in each bit plane and the statistical properties of adjacent pixels pairs in images is given. Then based on the relationship the method to estimate the embedding ratio of 2LSB is presented. The experimental results show that the proposed method has the advantage of higher accuracy compared with the existing methods when the secret message length of hyper–LSB and LSB are different, and it is also effective when the secret message length of hyper–LSB and LSB are the same.2) A steganalysis method which referred as EWS method for 2LSB is proposed. Firstly, the weighted stego–image model for 2LSB replacement steganography is constructed. Based on the model, a method to estimate the embedding ratio of 2LSB is brought forward. Experimental results show that the new method can estimate the embedding ratio in each bit plane of 2LSB effectively.3) Both the ERS method and the EWS method need the existing method of LSB steganalysis to estimate the rate of hyper-LSB. As an improvement, based on the model of the weighted stego–image and maximum principle, a steganalysis method for 2LSB replacement steganography is proposed, and the method can independently estimate the embedding ratio of 2LSB. The experimental results show that the novel method is effective, and the advantages and disadvantages of all three methods are given at the same time.Finally, the work of this thesis is concluded and some related issues and problems for future study are presented.