Enhanced stego-image quality and embedding capacity for the partial reversible data hiding scheme

Abstract

Recently, Jana et al. proposed the partial reversible data hiding (PRDH) based on (7, 4) Hamming code, which deals with three images: the original image (OI), the cover image (CI), and the stego image (SI). The CI is obtained by slightly modifying OI. After that, one can embed secret into CI to generate SI. The so-called partial reversible feature is that we can reconstruct the CI faultlessly, which is almost the same to the OI. However, in Jana et al.’s PRDH, the authors only adjust redundant bits by using (7, 4) Hamming code with even (or odd) parity. This simple approach may modify 3 redundant bits at most in 7 least significant bit (LSBs) to obtain CI from OI, and this degrades the CI quality seriously. In this paper, we construct two schemes: the proposed PRDH (PPRDH) and the modified PRDH (MPRDH). PPRDH enhances the partial reversible property that improves the visual quality of CI. And, MPRDH enhances PPRDH to achieve the high embedding capacity. Theoretical estimations of average mean square errors for these PRDH schemes are given to demonstrate the advantage of our PRDH schemes. In addition, we also point out two inaccurate descriptions in Jana et al.’s PRDH about position of embedded secret to more simplify the procedure of embedding secret.