Coverless Video Steganography based on Maximum DC Coefficients

Abstract: Coverless steganography has been a great interest in recent years, since it is a technology that can absolutely resist the detection of steganalysis by not modifying the carriers. However, most existing coverless steganography algorithms select images as carriers, and few studies are reported on coverless video steganography. In fact, video is a securer and more informative carrier. In this paper, a novel coverless video steganography algorithm based on maximum Direct Current (DC) coefficients is proposed. Firstly, a Gaussian distribution model of DC coefficients considering video coding process is built, which indicates that the distribution of changes for maximum DC coefficients in a block is more stable than the adjacent DC coefficients. Then, a novel hash sequence generation method based on the maximum DC coefficients is proposed. After that, the video index structure is established to speed up the efficiency of searching videos. In the process of information hiding, the secret information is converted into binary segments, and the video whose hash sequence equals to secret information segment is selected as the carrier according to the video index structure. Finally, all of the selected videos and auxiliary information are sent to the receiver. Especially, the subjective security of video carriers, the cost of auxiliary information and the robustness to video compression are considered for the first time in this paper. Experimental results and analysis show that the proposed algorithm performs better in terms of capacity, robustness, and security, compared with the state-of-the-art coverless steganography algorithms.