Processing of Hyperspectral Imagery for Contamination Detection in Urban Areas

Abstract

The application of hyperspectral imaging to detect soil contaminations is considered with respect to contamination assessment within urban areas. An optimal selection of hyperspectral imagery spectral bands is proposed. There is no reason to process all spectral bands as usually the 10-30 most informative ones are sufficient. A new criterion is introduced that incorporates such fundamental properties of hyperspectral imagery such as spatial resolution, mutual spectral information and signal-to-noise ratio. Algorithms for hyperspectral imagery processing and analysis were adapted for geochemical contamination detection. EO-1/Hyperion hyperspectral imagery was applied to central Kiev, Ukraine, and estimations of geochemical contamination of this area were mapped. © Springer Science+Business Media B.V. 2011.