Efficient Reversible Data Hiding Using Two-Dimensional Pixel Clustering

Pixel clustering is a technique of content-adaptive data embedding in the area high-performance reversible hiding (RDH). Using pixel clustering, pixels cover image can be classified into different groups based on single factor, which usually local complexity. Since finer seems to improve performance, this manuscript, we propose using two factors for two-dimensional develop RDH. Firstly, addition complexity, novel factor was designed as second clustering. Specifically, proposed defined rotation-invariant code derived from relationships four-neighborhood. Then, were allocated clusters factors, and cluster-based prediction realized. As result, prediction-error histograms (2D-PEHs) constructed, performance optimization selection expansion bins 2D-PEHs. Next, an algorithm fast expansion-bin introduced reduce time Lastly, realized according optimally selected bins. Extensive experiments show that significantly enhanced, particularly terms improved quality reduced capacity also moderately improved.