High-fidelity reversible data hiding based on pixel-value-ordering and pairwise prediction-error expansion

•This paper discusses the PVO embedding in two-dimensional (2D) space by integrating the pairwise prediction-error expansion, and a new 2D mapping adapted to the special distribution of pairs is proposed.•An adaptive 2D mapping selection mechanism is proposed to treat the rough and smooth pairs differently to further optimize the embedding performance. Pixel-value-ordering (PVO) technique refers to the process of first ranking the pixels in a block and then modifying the maximum/minimum for reversible data hiding (RDH). This paper discusses the PVO embedding in two-dimensional (2D) space and utilizes the prediction-error pair within a block for data embedding. We focus on not only the exploitation of conventional PVO embedding but also its effective implementation in 2D form. The PVO embedding is extended into a 2D form by integrating the pairwise prediction-error expansion, and a reversible 2D mapping adapted to the special distribution of prediction-error pairs is proposed. Moreover, an adaptive mapping selection mechanism is proposed to treat separately rough and smooth prediction-error pairs to further optimize the embedding performance. Experimental results show that the proposed method outperforms the previous PVO-based methods.