Reversible data hiding based on hybrid prediction and interleaving histogram modification with single seed pixel recovery

This paper proposes a reversible data hiding scheme for natural images. A hybrid prediction mechanism is utilized in order to produce prediction errors as many as possible. The cover image excluding a seed pixel is partitioned into four non-overlapping segments, and four predictors are tailored for each of them. As a result, most prediction errors concentrate around zero in prediction error histogram. Besides, an interleaving histogram modification mechanism is presented such that the capacity is enhanced and easier to be finely tuned in contrast to some previous approaches. Third, a single seed pixel recovery strategy is introduced. Experimental results show the effectiveness of the proposed method.