Effective reversible data hiding using dynamic neighboring pixels prediction based on prediction-error histogram

Abstract

In this paper, an efficient PEH-based scheme using dynamic neighboring pixels (DNP) predictor and adaptive embedding strategy is proposed. Concerning four gradient regions (horizontal/vertical/rhombus/plane gradient region) of the current pixel, the prediction method is generated adaptively by DNP predictor. After complexity calculating, the current pixel belonging to the flat, semi-flat and normal areas are applied for histogram modification and data embedding. Moreover, by using our selection strategy of thresholds and embedding bin pairs, the search procedure is accelerated and higher fidelity of stego-image is obtained. Experimental results indicate that our dynamic neighboring pixels prediction-error histogram (DNPPEH) scheme is more effective comparing with six state-of-the-art histogram-based RDH schemes in spatial domain.