A New Novel Gaussian Kernel-Induced Particle Swarm Optimization Based Spatial FCM Algorithm for Image Segmentation

Abstract

The advancement in image processing has witnessed massive progress in image segmentation technologies in medical as well as other fields. However, these advancements in technologies have not only made image analyst's life easy but also given them more challenging goals in image segmentation. Fuzzy c-means an unsupervised clustering algorithm plays a crucial role in the field of image segmentation. But suffer from many problems, such as center initialization, trapping into local optima and noise sensitivity. There are many algorithms which try to solve these problems. But no single approach is there which solves all these problems all alone. Hence, this paper proposed a novel Gaussian Kernel-induced PSO-based spatial FCM (GKPSOSFCM) which uses Gaussian Kernel Based Distance instead of Euclidean distance to increase efficiency, including spatial information to reduce noise sensitivity and Particle Swarm optimization that helps in finding global solutions. The result of the proposed method is compared with conventional Fuzzy C-means to demonstrate the efficiency of the proposed method. Experiments show that the proposed method outperforms the standard FCM algorithm and gives better-segmented results.