SPECTRAL SIMILARITY IN HYPERSPECTRAL IMAGE.

Hyperspectral imaging sensors measure the radiance of the materials within each pixel area at a very large number of contiguous spectral wavelength bands. The spectral information available in a hyperspectral image (cube) may serve to classify the nature of the target object because every material had a unique fixed spectrum. In this paper, the similarity can be described through various spectral distances between pixels and can be controlled by a threshold value. In the context of hyperspectral imagery, the “distance” is a spectral distance not a spatial distance. This measure is used in many classification techniques to estimate the distance between the spectrum to be classified and the representatives of the class („target”). The spectral screening is a process that can produce a representative subset in hyperspectral data with similar spectral signature. We have investigated the proposed method to find subsets of similar pixels on the AVIRIS data. [ABSTRACT FROM AUTHOR]