An analysis of spectral metrics for hyperspectral image processing

This paper investigates the efficiency of spectral metrics when used in spectral screening of hyperspectral imagery. Spectral screening is the technique of selecting from the data a subset of spectra such that any two spectra in the subset are dissimilar and, for any spectra in the original image cube, there is a similar spectra in the subset. The method can use various spectral metrics to characterize the similarity and can be seen as a data reduction step if the resulting subset is used in further computations instead of the full data. The investigation has focused on the comparison between spectral angle and spectral correlation angle in terms of efficiency of the results and speedup obtained as well as in empirically identifying the best distance threshold to be used when reducing the data. The techniques were tested on Hyperion imagery when using PCA and show promising speedup