Reversible data hiding for JPEG images with minimum additive distortion

Abstract

Reversible data hiding (RDH) in JPEG images aims at correctly extracting hidden data and correctly recovering the cover image. Existing methods for RDH, however, may not attain a good balance between good visual quality and small file size increment of the embedded image. To address the problem, we design a new RDH scheme by minimizing the additive distortion in JPEG images. Our method firstly designs different additive functions to calculate DCT block cost and DCT frequency cost, and then combines these two cost functions to minimize the additive distortion for a given embedding capacity. After obtaining the optimal embedding positions for each DCT block, we employ an improved two-dimensional histogram shifting strategy for data embedding, which significantly reduces the number of invalid DCT coefficient modifications. Extensive experiments demonstrate that our method outperforms existing JPEG RDH methods with better visual quality and smaller file size increment of the marked image.