Genetic Algorithm Using Feistel and Genetic Operator Acting at the Bit Level for Images Encryption

Abstract

In this paper, a new medical image encryption technique based on genetic algorithms acting at the bit level will be developed.Initially, a transformation to a binary matrix notation of the original image is applied, followed by an evaluation function determined by the Hamming distance between the obtained image and another pseudo-random image generated from chaotic maps used.This discrimination function divides the image, viewed as a population where each row represents an individual, into two categories: a strong population and a weak population.An enhanced Feistel round will be implemented by introducing a chaotic mating between the two categories based on a circular shift for the right bloc and a pseudo-random permutation for the left bloc.Next, a genetic crossover adapted for image encryption will be performed with another pseudo-random vector under the control of a crossover table.To ensure the robustness of our approach, a genetic mutation will be applied at the end of the encryption.A multitude of images of different sizes and formats have been tested using our approach, with encouraging results.