Binary coding of imaging spectrometer data for fast spectral matching and classification

Abstract Several simple binary codes are applied to AVIRIS data to develop a spectral representation that facilitates efficient library searching in applications where identification is dependent on matching a measured spectrum against library prototypes. Excellent results are demonstrated using libraries generated by clustering image segments, suggesting the value of the procedure in general. Drawing on this experience, it is shown that good thematic mapping can also be achieved with spectra that have been binary-coded, using algorithms based on minimum Hamming distance measures.