A reversible data hiding using difference-histogram modification on multi-directional in two-dimensional histogram

Abstract

A new approach to Reversible Data Hiding (RDH) is introduced in the proposed work named as Difference-Histogram Modification on Multi-Directional (DHMMD). This method achieves better embedding capacity and Peak Signal-to-Noise Ratio (PSNR) compared with the existing methods. Recently, Li proposed a method which embeds the data in four directions (i.e., left, right, up, and down), the proposed work enhanced the Li’s method with one more direction (i.e., diagonally up and down) for embedding. This modification leads to get more pixel pairs for embedding and less for shifting. This eventually increases the embedding capacity. The overhead data (information about embedded data and LSB bits) will be embedded in watermarked image itself using Least Significant Bit (LSB) replacement. Gradient Adjusted Prediction (GAP) and noise level were adopted for finding the priority pixel pairs for embedding. This method achieves a better PSNR value more than 50 dB in majority of images from USC-SIPI database when compared with earlier RDH methods.