Reversible Data hiding based on Three-Level Hierarchical Embedding with Histogram Mapping in Encrypted Images

This paper proposes a reversible data-hiding (RDH) approach based on a three-level hierarchical embedding with histogram mapping in encrypted images for a cloud environment. The proposed approach uses two-level encryption followed by the three-level hierarchical embedding process. The two-level encryption uses an intra-inter block permutation with an exclusive OR operation. In the three-level hierarchical embedding two predictors namely strong predictor and weak predictor are estimated. The strong predictor is used in level 1 embedding, while the weak predictor is used in level 2 embedding. A histogram mapping is then performed on the location map obtained during the level 1 and level 2 embedding process and is then compressed using the Huffman coding. The compressed location map information is then further embedded as the level 3 embedding process. In the extraction phase, the compressed location map information is first extracted, then the level 2 data followed by the level 1 data. The evaluation of the proposed data embedding approach is done using the Bossbase and Bows-2 datasets that result in an embedding rate of 3.806 bpp and 3.513 bpp respectively.