Morpho-spectral objects classification by hyperspectral airborne imagery

Abstract

Cities are characterized by a complex mosaic of objects, representing the urban structures, the history and the transformations. The characterization of urban objects requires powerful methods combined with high resolution imagery, in this study we present an object characterization method that takes into consideration the spatial and spectral characteristics of remote sensing imagery, using an airborne hyperspectral image. The method consists of two mains steps; 1) a spectral classification of the objects using an external spectral library combined with image collected spectra, 2) a morphological classification of the objects using their geometric attributes. The goal is to provide an efficient objects characterization method that takes advantage of both spatial and spectral dimensions of hyperspectral imagery, and to improve classification methods efficiency.