Reversible data hiding with pairwise PEE and 2D-PEH decomposition

•We propose a novel RDH with pairwise PEE and 2D-PEH decomposition.•The proposed method can achieve good capacity-distortion performance.•The proposed method outperforms some state-of-the-art RDH methods. Reversible data hiding (RDH) is an important topic of data hiding. In this paper, we exploit pixel-based pixel value ordering prediction (PPVO) and pairwise prediction-error expansion (PEE) to design a novel RDH with pairwise PEE and 2-dimensional prediction-error histogram (2D-PEH) decomposition. Specifically, for each pair of pixels, they are accurately predicted according to the maximum and minimum values of their neighbor pixels by using PPVO to generate a global 2D-PEH. After selecting negative 2D expansion bins, a series of sub 2D-PEHs can be obtained by decomposing the global 2D-PEH according to the errors between the maximum and minimum values of neighbor pixels. Finally, for each sub 2D-PEH, pairwise PEE mapping manner is selected to conduct data embedding according to the characteristic of the sub 2D-PEH. Experimental results demonstrate that the proposed method can achieve good capacity-distortion performance and outperforms some state-of-the-art RDH methods.