Feature extraction via spectro-temporal analysis of hyperspectral data for vegetative target detection

Abstract

In this paper, the authors investigate the use of hyperspectral-multitemporal features for discriminating between two aquatic weed species, Waterhyacinth and Bulrush. Hyperspectral, multitemporal data is three- dimensional data that can be organized into a "spectro- temporal map" where the x-axis is time, y-axis is wavelength, and z-axis is reflectance. The authors present an algorithm based on a greedy search approach to extract pertinent features to solve the classification problem at hand.