On content-based image retrieval systems for hyperspectral remote sensing images

Abstract

This chapter includes a review of some key elements of Content Based Image Retrieval Systems (CBIR). We place into this context the the works found in the literature regarding remote sensing images. Our own focus is on hyperspectral images. The approach we are pursuing is that of characterizing the spectral content of the image through the set of endmembers induced from it. We describe some ideas and numerical experiments for a system that would perform CBIR on hyperspectral remote sensing images. We propose a spectral image similarity to guide the search to answer the queries. © 2008 Springer-Verlag Berlin Heidelberg.