Pixel Value Ordering Based Prediction for High-Fidelity Images
Ravi Uyyala, Padmavathi Vurubindi, M Anila
Abstract
Pixel Value Ordering is a method of Reversible data hiding (RDH) where the pixels in a block are arranged in a proper order for predicting the single pixel value. The secret data bits will be embedded based on the highest or lowest value within the blocks. This work presents a PVO method using Quad-tree decomposition for RDH. Initially, the Quad-Tree decomposition technique has been used for decomposing the image into various uniform blocks which are optimized for better embedding. For a better understanding of the image pixel values pair-wise prediction error technique has been proposed in this work. A 2-D mapping scheme has been proposed for better correlation among the prediction errors. The suggested strategy performs better than previous RDH techniques, according to experimental results.