Reversible data hiding using additive prediction-error expansion

Abstract

Reversible data hiding is a technique that embeds secret data into cover media through an invertible process. In this paper, we propose a reversible data hiding scheme that can embed a large amount of secret data into image with imperceptible modifications. The prediction-error, difference between pixel value and its predicted value, is used to embed a bit '1' or '0' by expanding it additively or leaving it unchanged. Low distortion is guaranteed by limiting pixel change to 1 and averting possible pixel over/underflow; high pure capacity is achieved by adopting effective predictors to greatly exploit pixel correlation and avoiding large overhead like location map. Experimental results demonstrate that the proposed scheme provides competitive performances compared with other state-of-the-art schemes.