A Feistel Network Followed by a Bitwise Crossover for Image Encryption

This document proposes an innovative approach to color image encryption, based on an enhanced version of the Feistel scheme. This approach utilizes the widely employed chaotic maps within the realm of cryptography, notably the logistic map and the PWLCM map, to operate at the binary level on blocks of pseudo-random dimensions. Following this step, a genetically designed crossover process specifically tailored for image encryption is applied to the resulting vector, supervised by a random decision table. Subsequently, the result of the Feistel scheme is divided into 3m sub-vectors, each with a size of (1; 8n), and each block is confused with a pseudo-random block of the same size, depending on the values of the decision table. Simulations conducted by our algorithm on many randomly selected images from a database have revealed highly promising statistical and differential results, thus ensuring the robustness of our cryptography system.