ALGORITHMS FOR PROCESSING MULTISPECTRAL IMAGES WITH CERTAIN INVARIANT PROPERTIES

Abstract

Algorithms for processing multispectral scanner images are described, the results of which are not affected by additive and multiplicative components of error in brightness measurement, which are unknown but identical in all spectral channels. A detailed description is provided of an algorithm for the joint processing of images recorded in three spectral channels, but the approach can be adapted for use in processing imagery in four, five, or more bands. Classification maps produced by the algorithm for the purpose of discrimination of water, bare or sparsely vegetated ground, and areas of denser vegetation are presented and assessed.