A Secret Message Extraction Algorithm And Steganography Algorithm Design

With the great development of communication, Internet becomes very important in our life. For the safety of the data in internet, people embed it in some open data media to make others ignore the existence of the data. After reading various steganography algorithms and steganalysis algorithms, we found many steganalysis algorithms emphasized on the existence of secret message in the present while seldom referred to extracting secret message, because extracting secret message is very hard without the key. In this condition, we focused on an approach to extracting hidden message in an especial instance and pay attention to the design of steganography scheme.At first, a steganalysis approach of extracting hidden message is proposed. The approach aims at some steganographic algorithms of spatial LSB insertion in the conditions of cover-object used repeatedly. Material thought is: We get some different points by comparing two stego-objects from one cover-object, and then utilize contaminated distribution and hypothesis testing to sort the points, next, cryptanalysis is combined to look for the stego-key, in the end, we have elaborate analysis of it. Results of experiment to "hide and seek" are provided to prove that stego key can be found effectively.Second, a new steganography algorithm is proposed to the steganalysis approach of extracting hidden message above. A m-sequence is used to control the embedding of secret message so that the highest significant bit, secondary significant bit and the least significant bit can join the embedding process together. Since we only embed message in the point of different value in the highest significant bit and secondary significant bit, if the m-sequence is unaware, the algorithm can resist cover-attacks triumphantly.At last, another steganography algorithm is proposed owing to research many steganography algorithms and steganalysis algorithms previously. Image divided into blocks and selecting keys previously are used in embedding process, at the same time we change the value of pels by plus 1 or minus 1 according to adjacent pelses. Selecting keys previously makes fewer modifications, and image divided into blocks guarantee secret message in the appropriate pels. While changing the value of pels by plus 1 or minus 1 break the durance of the LSB-based algorithm, which only does 2% 2i +1, so that the algorithm can resist RS. Results of experiment also show that this method can effectively resist RS, and it's quality is more ascendant than others.