Supervised images classification using metaheuristics
Amir Mokhtar Hannane, Hadria Fizazi
COMPUTER MODELLING & NEW TECHNOLOGIES 2016 20(3) 17-23
Department of computer science, University of sciences and technology of Oran (Mohamed Boudiaf), BP. 1505 El-Mnaouer 31000, Oran, Algeria
Image classification is a fundamental task in image processing because it is a crucial step toward image understanding. This paper exploits metaheuristics (Ant Colony Optimization and Electromagnetic Metaheuristic) to tackle the problem of supervised satellite image classification. Earlier studies have been used the Intra-Class Variance (ICV) for images classification but this function has a limits to solve classification problem. This study presents the introduction of the Davies-Bouldin Index (DBI) to the supervised images classification. This index is used in two stages: training step and classification step. In training step this index serve as criteria for controlling iterations. In the classification step this index help to classify each pixel in the image to their appropriate class using the class centers found during the training stage. The experimental results show that the introduction of the Davies-Boulin index is very effective for supervised images classification and help the community of researches to improve the classification accuracy of remotely sensed data. The utility of metaheuristics is also demonstrated for satellite image of Oran city.