Using Gaussian synapse ANNs for hyperspectral image segmentation and endmember extraction

Abstract

Remote sensing of the earth is becoming a commonplace technology which is growing driven by the combined advances in sensor technologies and image and data processing. In the near future the international proliferation of remote sensing devices will require an ever increasing number of high-resolution systems with lightweight instruments for application in small satellites and light planes, as well as a reduction of costs. One important tool used in remote sensing is imaging spectroscopy, also known as multi, hyper or ultraspectral remote sensing. It consists in the acquisition of images where for each spatial resolution element a part of the electromagnetic spectrum sampled at different rates is measured [1,2]. © 2008 Springer-Verlag Berlin Heidelberg.