Feature selection approach of hyperspectral image using GSA-FODPSO-SVM

The aim of this paper is to classify the object in hyper spectral images which are high dimensional images and consists of many data channels. Another aim is to use machine learning classification algorithm like support vector machine (SVM) which is good for high dimensional data case. SVM provides a good accuracy of classification. A statistical model is developed to learn and classify hyper spectral data using the low dimensional representation. For this purpose we used a combination of evolutionary optimisation algorithms which are GSA (gravitational search algorithm) and FODPSO (finite order Darwinian particle swarm optimisation).