Integrating multi-predictions encryption with histogram shifting secret-sharing for high-capacity two-layer image data hiding

Abstract

ABSTRACT With the aim to increase the embedding capacity as well as the quality of reversible data hiding (RDH), we present a high-capacity two-layer RDH based on multi-predictions interpolation and histogram shifting taking advantage of encryption and secret sharing. The highlight of this paper is to improve IRDH (Interpolation-based RDH) using secret sharing and multi-predictions. Secret sharing is designed for full ownership and partial ownership. Multi-predictions including Cross-Shape Average Prediction Interpolation (CSAPI), Weighted Prediction Interpolation based on Euclidean Distance (WPIED) and Different Direction Gradient Prediction Interpolation (DDGPI) are firstly proposed to hide data in different blocks in layer-1. Different blocks are divided by five texture features of EECVM (Energy-Entropy-Contrast-Variance-Mean). In layer-2, the difference histogram shifting method based on eight modes is adopted. Furthermore, ‘Matrix Traversal and Snake Scrambling’ and ‘Cyclic-shifting and S-box encryption’ are proposed for improving the security in watermarking images and carrier images. The experimental results provide superior outcomes with high capacity, imperceptibility, along with a good security.