Effects of the spatial enhancement of hyperspectral images on the distribution of spectral classes

In this chapter, we present a study on the effects of the spatial enhancement of hyperspectral (HS) images on the distribution of spectral classes. The analysis is based on the concept of dimensionality reduction, the transformation of high-dimensional data into a meaningful representation of reduced dimensionality which may favor visualization and understanding of high-dimensional data. Nonlinear techniques of dimensionality reduction are applied to original Hyperion HS data (30 m) and to fusion products with the panchromatic channel of ALI (10 m) obtained from different sharpening methods, in order to evaluate possible advantages or critical situations deriving from multi-sensor, multi-resolution data fusion.