Land Classification Based on Hyper Spectral Images using Deep Learning Techniques

Abstract

The study of chemical and physical properties of a remote sensing data is done by one of the form called as Hyper Spectral image. The Hyper Spectral image (HSI) is a captured data with consistent materials in a nonlinear relation form. Each HSI has specific wavelength with spectral reflectance in a matching entries on vector with high dimensional pixels. Although classification of HIS performanceis good based on spectral-spatial but they depend heavily on hand craft or based on shallow descriptors. The ability of representing features in the form of custom made is not sufficient to label the dissimilarity among the classes of altered or same.Extracting the features is measured as essentialtechnique in HSI classification. To extract the features Deep Learning method is used due to classifying the 2D and 3D dimensions and to extract certain shapes in an image etc., can do clearly. And compared what outcomes will come by applying deep learning to the data using Big Data.