An adaptive PEE-based reversible data hiding scheme exploiting referential prediction-errors
Fei Peng, Xiaolong Li, Bin Yang · 2015
Prediction-error expansion (PEE) is an efficient technique for reversible data hiding (RDH). Instead of expanding the highest histogram bins in conventional PEE, in this paper, to better utilize the image redundancy, we propose a new PEE-based RDH scheme with an advisable expansion strategy utilizing referential prediction-errors. For each pixel, we first calculate its prediction-error and use its neighbor prediction-error as a reference. The correlation of the prediction-error and its reference is exploited to adaptively select bins for expansion embedding. In addition, to further enhance the reversible embedding performance, we apply the pixel selection technique in our scheme such that the pixels located in smooth image areas are priorly embedded. Experimental results show that the proposed scheme outperforms conventional PEE and it is better than some state-of-the-art RDH works as well.