An Adaptive Reversible Data Hiding Scheme Using Two- Dimensional Histogram Modification

Reversible data hiding (RDH) has witnessed significant progress and utilized applications across various domains, including medical imaging, military, and cloud computing. This paper proposes an adaptive RDH based on a novel two-dimensional (2-D) prediction error expansion (PEE) and improved pixel value ordering (I-PVO) mapping to enable a useful trade-off between embedding capacity (EC) and image distortion. Using this scheme, the cover image is first processed using a sliding window size of 4 × 4. Then, each window is divided into inner and outer sub-blocks. The inner sub-block statistical characteristics are determined using the standard deviation of the pixels in the outer sub-block. If this standard deviation is below the first specified threshold, a novel 2-D mapping of PEE strategy is employed for data embedding. Conversely, if the standard deviation falls between the first and the second given thresholds, then conventional pairwise I-PVO is used. Blocks that do not satisfy either condition are bypassed, and no secret data is embedded. This adaptive approach allows the RDH scheme to optimally use both PEE and I-PVO techniques, resulting in higher EC and improved image quality compared to previous RDH methods.