A Novel Cryptosystem Using Bitwise Permutations on Rows and Columns Followed by a Pseudo-Random Adapted Genetic Crossover

Abstract

Most image encryption algorithms operate at the pixel level, treating it as the primary element. However, bitwise permutations offer greater efficiency by altering both the pixel's value and its position within the image. In this study, we propose a novel color image encryption technique that applies multiple independent permutations to the rows and columns of the image after a Pseudo-Random transition to binary notation. Following a Pseudo-Random return to the grayscale level, a genetic crossover highly suited for color image encryption is applied to the resulting vector, integrating a substitution process using two S-Boxes and a diffusion mechanism to counter differential attacks. Extensive simulations on images of various sizes and formats confirm the robustness of our method against known attacks.