Hyper spectral image classification and unfixing by using ART and SUNSPI techniques

Abstract

The Hyperspectral images extracts, collects and processes the information from across the electromagnetic spectrum. The main aim of the hyperspectral imaging is to get the spectrum from each pixel in the images, in the purpose to find the objects, materials, or detecting processes. The spectral range in the hyperspectral images gives the ability to identify chemical types on the environment of Mars more precisely than before. For extracting the hidden features in the mixed pixel, demonstrating state-of-the-art presentation when evaluate with freshly established hyper spectral image classification techniques. Then proposed method is experimentally calculated by using both pretended and actual hyperspectral datasets. The integration of Unmixing algorithm termed “Sparse Unmixing of Hyperspectral information with Spectral a Priori data” with the Singular Spectrum Analysis approach, to get the better result the Clustering by “Adaptively Regularized Kernel-Based Fuzzy C-Means” and Segmentation with “Watershed” of images is carried out and for the better level of classification is done using the ART classifier. The integration of these methods signifies an innovative contribution in the research field of hyperspectral imagery.