Restauration Adaptative d’image par les méthodes intelligentes

Abstract

In this work we tried the application of evolutionary techniques on image restoration field, specially the Particles Swarm Optimization (PSO) which was used in different ways. First, we have used the basic PSO algorithm in two manners: in local restoration, where the image constitutes the entire swarm and pixels are its particles, and global restoration where the image is considered as particle and a swarm is built by similar ones. We have obtained good results compared with other developed image restoration techniques. As an evolutionary technique, the PSO can suffer from the initialization phase, so we have proposed a solution. In this solution we introduced the Search Efficiency Function (SEF), based on Levy flight, as initialization step of the PSO. The obtained results were excellent which prove our right choice and contribution. Also the introduction of this function, SEF, has allowed the execution of a blind restoration and the results were surprising