High-fidelity reversible data hiding by Quadtree-based pixel value ordering

Recently, the pixel value ordering (PVO) method has received a great deal of attention in the field of high-fidelity reversible data hiding. To improve the embedding performance of the PVO-based method, a quadtree-based pixel value ordering (QPVO) method based on dynamic quadtree partition is proposed in this paper. Instead of using equal-sized blocks, blocks in various sizes are adaptively generated based on the quadtree partitioning and block complexity. Smooth regions are divided into smaller blocks to obtain high embedding capacity, while rough regions are divided into larger blocks to avoid distortion. The decoder can recover the original image as well as the original quadtree structure by the block complexity. Extensive experiments demonstrate that the proposed QPVO method could significantly improve the embedding performance of these PVO-based methods, especially for a relatively small embedding payload.