High capacity reversible data hiding using overlapping difference expansion

Difference expansion (DE) has been widely used for reversible data hiding. In this work, a new DE based scheme is presented that uses consecutive, overlapping pairs, instead of the non-overlapping pairs or triads used by traditional DE derivatives. The scheme is superior to the existing approaches, both in capacity and PSNR terms. By applying multiple runs of the embedding process, a significant capacity gain is obtained at the expense of lower quality.