Efficient reversible data hiding algorithm based on gradient-based edge direction prediction

In this paper, we present an efficient RDH algorithm based on a new gradient-based edge direction prediction (GEDP) scheme. Since the proposed GEDP scheme can generate more accurate prediction results, the prediction errors tend to form a sharper Laplacian distribution. Therefore, the proposed algorithm can guarantee larger embedding capacity and produce better quality of marked images. The determination of appropriate thresholds is also a critical issue for a RDH algorithm, so we design a new systematic way to tackle this problem. In addition, a modified embedding order determination strategy is presented to reduce the distortion of a marked image. Based on typical test images, experimental results demonstrate the superior properties of the proposed algorithm in terms of embedding capacity and marked image quality. (c) 2012 Elsevier Inc. All rights reserved.