Current Proceedings on Technology
Yazarlar: Tahir Sağ, Mehmet Cunkas
Konular:-
Anahtar Kelimeler:Image segmentation,Multi-objective optimization,Clustering,Swarm intelligence,PSO,NSPSO
Özet: In this study, a multi-objective particle swarm optimization algorithm is proposed for image segmentation by means of clustering. The proposed algorithm, called non-dominated sorting PSO (NSPSO), has been developed by adapting the running principle of NSGA2 algorithm that is widely used through multi-objective evolutionary algorithms. NSPSO algorithm that is adapted to image segmentation and also three single objective clustering algorithms are applied to three test images. These are FCM and K-Means, which are well known classic clustering algorithms, and simple PSO. Same parameter values are used for running each of these algorithms 20 times. With respect to experimental results, the proposed approach employed multi-objective optimization is more successful than single objective techniques.