Research On A Multi-model Information Hiding System

In the environment of open Internet and fast-developing big data technique,there are potential security issues with the transmission of information in the network.The spread of special information on the Internet often involves the interests of individuals,society and nation.Therefore,it is of great importance to implement covert communication,while information hiding is a vital method towards this issue.Many steganography algorithms have emerged in the past 20 years of development;meanwhile,new steganographic and steganalysis algorithms are constantly being updated,as well as the mode and framework of information hiding.However,it is of urgent need to improve the practicability of most existed methods.Thus,this paper proposes a coverless information hiding algorithm and modifies two embedding steganography algorithms to design a multi-modal information hiding system by leveraging the multi-model method of information hiding.The main work of this paper is as follows:1.A spatial domain-based and a transform domain-based steganographic algorithm are modified.We analyzed the traditional steganography methods,which allow the cover image to be modified in accordance with the embedding rules,pointing out that 2-bit-binary digits can be respectively embedded into different color bands corresponding to the same points by using the complementary embedding rule.We modified LSB matching algorithm,allowing secret information to be embedded into the complex texture areas.In which,the gray level co-occurrence matrix is leveraged to analyze texture characteristics;the target embedded sub-block is filtered by the threshold setting,while the secret information is embedded into the color bands of target blocks with complementary embedding rules.Moreover,we modified the two-point DCT embedding algorithm.In which,the 2 binary data and the DCT coefficient matrix work together,taking control of the message embedding by using algebraic logic relations.It averagely changes 0.5 DCT coefficients while embedding 2 bit data,which is of higher embedding efficiency compared to other DCT-based methods.2.A coverless information hiding algorithm based on grayscale gradient co-occurrence matrix is proposed.We deeply summarized the basic principles of coverless information hiding methods;analyzed the 3D map of the grayscale gradient co-occurrence matrix of images,and observed the changing rules and trends of the image points;removed elements of high gradient level to improve the resistance of low-pass filtering;converted the grayscale gradient co-occurrence matrix of an image into a 7-bit-binary sequence by projection,averaging and quantization.This algorithm shows great robustness to from low-pass filtering and salt&pepper noise attack.3.A multi-model information hiding system with jointed algorithms is designed.In order to improve the diversity of the system and increase the difficulty of steganalysis from third parties,two different steganography frameworks and three algorithms are combined to design and implement a multi-modal information hiding system.In which,the stego keys are responsible of the algorithms selection,and controlling the stitching orders of the images which generated under coverless information hiding method.