High-fidelity lossless data hiding based on predictors selection

Abstract

Reversible data hiding based on prediction methods is a data hiding technique wherein secret bits can be efficiently hidden into cover images. In this paper, we propose a reversible data hiding method based on multiple prediction methods and local complexity. At each level of data hiding algorithm, we evaluate four prediction methods to decide which method should be chosen to embed secret messages. We propose two tactics to evaluate and select prediction methods. When a prediction method is chosen to perform a shifting and embedding process, a threshold based on local complexity is used to determine which pixel should join the shifting and embedding process. If the local complexity of a pixel is smaller than the threshold, the pixel will join the process; otherwise, the pixel will cease to join the process. Therefore, more pixels will avoid executing pixel shifting. Doing so results in stego-images with lower distortion. The experimental results show that our embedding capacity and quality is superior to those of other approaches.