“…30 and 37, for a particular value of T as well as a particular value of N , mn being constant for an m × n image. In state-of-the-art pixel-prediction based reversible watermarking algorithms [13,14,16], usually one pixel in each 2 × 2 prediction block represents the base pixel, hence about 3 4 th of the total number of pixel locations are selected for embedding. So, in the proposed framework, we take N mn = 0.75.…”