Hyper Spectral Image Classification using Natural Computing Strategies

Abstract

Hyper-Spectral Image (HSI) classification has become an intriguing era for researchers with regards to late years due to its different applications like Land Use/Land Cover, water resource management, surveying and urban planning and many more. This paper includes cogent Natural Computing strategies analogies for Hyper Spectral Image (HSI) Classification. HSI have various issues related to it because of its size, which decreases the robustness and performance of classifiers. The paper contemplates to give emphasis on point that Natural Computing strategies are better optimizer for Terrain feature extraction from satellite images. In this paper, three natural computing techniques are used in hybridized form. As hybridization yields better results by combining the good features of these algorithms with each other; one of which is Particle swarm Optimization (PSO) with Artificial Neural Network (ANN) and the other is Particle Swarm Optimization (PSO) with Artificial Neural Network (ANN) and with some feature of Genetic Algorithm (GA). The combined form of algorithms is used to combine the advantages of each technique so that better accuracy can be achieved for high land cover feature from Hyper Spectral Image. To test the accuracies of these hybridized techniques three hyper-spectral satellite dataset are used.