Extraction and selection of robust features for classification of multispectral remote-sensing images

Abstract

In this paper, we present an approach to the extraction and selection of robust features for classifEation of multispectral remote-sensing images. In particular, several robust features are proposed that, given a specific landcover class, aim to exhibit an invariant behavior versus variations in the acquisition conditions of the images considered. In addition, a technique is presented, which is able to adaptively select the most robust features for a given problem. 0-7803-5207-6/99/$10.00 0 1999 IEEE.