DYNAMIC RANGE OPTIMIZATION FOR REVERSIBLE DATA HIDING WITH INTENSITY-BASED EMBEDDING

Abstract

Reversible data hiding is a technique that enables the recovery of the original media after the hidden data has been extracted. This feature is crucial in scenarios where any permanent alteration to the cover media is unacceptable. Examples of such applications include law forensics, satellite imagery, medical imaging, and cloud data management. In these fields, preserving the integrity of the original media is essential for accurate analysis and decision-making. Current advanced RDH approaches typically embed extra information into the cover media without taking pixel intensities into account. This uniform embedding strategy often leads to poor visual quality in the modified media, as the process does not differentiate between regions of varying sensitivity or intensity within the image.

In this work, the dynamic range of cover images is utilized to embed additional data based on pixel intensities. The proposed method involves dividing the cover image into two segments: part A and part B. Part A serves as the area for reversible data embedding, while part B is designated for storing the least significant bits of the pixels in part A.

The LSBs in part A are vacated according to predefined ranges that align with the human visual system and Weber's law. These extracted LSBs are then reversibly stored in part B using a histogram shifting-based RDH technique. This structured approach ensures that the embedding process is both reversible and minimally intrusive.

Upon reception, the process begins with the extraction of the additional data from part A. Subsequently, the hidden LSB stream in part B is retrieved, enabling the restoration of the original pixels in part A. This two-step extraction guarantees the reversibility of the embedding process.

Experimental evaluations conducted on a publicly available image dataset reveal that the proposed RDH methods achieve high embedding rates while maintaining low distortion in the modified images. Furthermore, the new approach demonstrates superior performance compared to existing state-of-the-art RDH techniques.

Keywords : Reversible data hiding, pixel intensity, dynamic range, HVS, Weber's law