HSB based reversible data hiding using sorting and pairwise expansion

Abstract

Reversible data hiding (RDH) continues to advance in its applicability across diverse fields, including medical imaging, military applications, and cloud computing. In recent times, there has been a proliferation of RDH methods that focus on embedding in the higher significant bit (HSB) plane, as opposed to the least significant bit (LSB), in order to leverage the greater correlation, thereby achieving a larger embedding capacity (EC). This paper introduces a novel and efficient RDH approach that utilizes sorting and pairwise expansion. The proposed method initiates by partitioning the original image into HSB and LSB planes and then converting the HSB plane into a chessboard pattern. Subsequently, pixels are organized in ascending order based on their predicted values, forming groups of 1 × 3 pixels. The embedding process occurs in two passes: the first employs a difference expansion-based RDH, while the second pass employs pairwise expansion to minimize distortion. This method accomplishes an expanded EC while preserving the quality of the resulting image. Experimental results also demonstrate that the proposed approach maintains a significant lead of approximately 0.5 dB over the best-performing existing methods, underscoring its superior embedding efficiency.