Abstract :
A feature selection criterion is proposed which is a hybrid of a generalized interset and intraset distance. The problem considered is that of selecting the G best wavebands from a set of H possible wavebands (G<H). The selected wavebands can then be used as inputs to a small computer for decision processing.
Keywords :
Feature selection, interset distance, intraset distance, multispectral, nonparametric, pattern classification.; Application software; Cost function; Covariance matrix; Degradation; Electromagnetic measurements; Electromagnetic spectrum; Error analysis; Pattern classification; Size measurement; Time measurement; Feature selection, interset distance, intraset distance, multispectral, nonparametric, pattern classification.;
