Histogram-based reversible data hiding technique using subsampling

Abstract

Reversible data hiding technique guarantees that the original cover image is completely recovered from the marked image after the hidden data are extracted. This technique can be applied in some applications such as medical imaging and law enforcement where any distortion to the original image is not acceptable. In this paper, we present a reversible data hiding algorithm that shifts the histogram of the differences between subimages obtained through subsampling and embeds secret data by modifying the pixel value. Since the subimages are highly correlated and have strong spatial correlations, it utilizes the bins whose differences have smaller magnitude in the difference histogram. Moreover, the proposed scheme prevents possible over/underflow problem and no location map is needed during the retrieval. Experimental results prove that the proposed algorithm provides high embedding capacity while keeping the distortions low. In order to demonstrate the validity of our algorithm, we compare to other histogram-based reversible data hiding algorithms in terms of capacity and visual quality.