Evaluation of vegetation indices with optimal bands for vegetation classification using CASI Hyperspectral Image

Spectral indices are combinations of surface reflectance at two or more bands which indicate spectral characteristics for a particular material on ground. Many spectral indices including Normalized Difference Vegetation Index (NDVI), which is a widely used spectral index to analysis vegetation area using spectral characteristics of vegetation on red and near infrared wavelength ranges, are studied using multispectral images for vegetation analysis. As hyperspectral image with precise spectral information in hundreds of bands appeared, some spectral indices were applied on hyperspectral images. However, studies of spectral indices for hyperspectral images using optimal bands are insufficient on the analysis of various vegetation targets comparing to the studies for multispectral images. In this study, we have selected spectral bands from CASI image and applied them to study data. We performed the classification using the vegetation indices calculated with the selected hyperspectral bands to evaluate discrimination results of three different crops for classification.