Efficient Matrix Embedding Based on Random Linear Codes

Matrix embedding has been proved to be an important steganographic scheme that can improve the embedding efficiency. In this paper, we propose a novel matrix embedding scheme with improved embedding efficiency. The original optimal embedding problem was converted to a problem of finding an optimal solution for over-defined systems of equations. Walsh spectrum and fast Hadamard transform are used to search for the optimal solution with reduced computational complexity. Moreover, a method of bit control is proposed to further increase the embedding efficiency. Based on the proposed scheme, a family of efficient embedding codes is given by using GZZW construction.