Performance evaluation of urban areas Land Use classification from Hyperspectral data by using Mahalanobis classifier

A growing urbanization required continues observation of Land Use Land Cover (LULC). A Geospatial Technology can provide a sustainable LULC with a good accuracy as compared to traditional way. From past few decades a Panchromatic and Multispectral data were widely used; now with advent in Geospatial Technology a Hyperspectral data can be available for it. Hyperspectral imagery contains diverse information from a wide range of wavelengths. Due to the mix-structure of an urban area, it is very difficult to identify and the classification of an objects. For this work EO-1 Hyperion imaging data were used. Then layer stacking was performed in ENVI tool, after this data was converted to Band Interleaved by Line (BIL) format. Hyperion data has 242 bands but few bands were identified as Bad bands, by removing these bad bands only 196 bands were considered for making Hypercube. Then Mahalanobis classifier was applied and the accuracy of classifier was 88.46% with Kappa Coefficient 0.84.