A square lattice oriented reversible information hiding scheme with reversibility and adaptivity for dual images

This paper proposes an adaptive dual-image-based reversible information hiding scheme. First, for a pixel pair in the original image, an associated square lattice is selected to determine the maximum distortion that will be induced by hiding information. Then, an embedding rule, which combines the selected square lattice with integer rounding, is constructed. In addition, a novel histogram shifting mechanism to solve the problems of overflow and underflow is designed to decrease the distortions that are caused by the shifting of a large number of pixels. In our approach, both the original image and the secret messages can be recovered losslessly. Our experimental results show that the proposed scheme achieved performance that was superior to the state-of-the-art schemes, resulting in embedding rates up to 1.40 bpp and average PSNRs of approximately 47.60 dB. The proposed scheme also can be used in real-time systems due to the simplicity of its computations.