Fractional order Darwinian particle swarm optimization based segmentation of hyperspectral images

T. Arul Kumaran and U. M. Cami

Abstract

Hyper spectral images are of high dimension. There are many number of data channels in a hyper spectral image. Segmentation of hyper spectral images is very difficult. In this paper a new segmentation technique for multispectral images is proposed. This paper introduces a concept that combined algorithm of FCM (fuzzy C) and fractional order Darwinian PSO can perform better in terms of classification accuracy. Fractional-order Darwinian particle swarm optimization (FODPSO) uses many sets of test data. Junction rate of particles are controlled by use of fractional derivative concept. Otsu problem is solved using this concept in remote sensing data. This paper classifies various features that are related to any remote sensing hyper spectral image. These features help us to analyse the images better for using in various applications.