Developing Reversible Data Hiding with Interpolation and Local Dispersion

Recent Interpolation-based Reversible Data Hiding (IRDH) schemes demonstrated significantly higher embedding capacity, however, their embedded image quality is not much impressive, particularly for higher size payload. This paper presents an improved IRDH scheme with local dispersion for better quality embedded image. For high embedding capacity, correlation based embedding is utilized in all embeddable bits only in the interpolated pixels. The error between an embedded pixel and its original version is then reduced using the local (i.e., block-wise) standard deviation to further improve the embedded image quality. Since the standard deviation is computed from the original pixels and their respective interpolated pixels in a block, the improved embedding process remains reversible to losslessly extract the embedded data. All the original pixels are also kept intact for perfect recovery of the original image. A significant improvement in embedded image quality is thus recorded in comparison to the state-of-the-art IRDH schemes.