Hyperspectral mosaicking, coregistration and change detection

This paper describes an integrated pair of tools, Geobox and Hawkeye, that facilitate quantitative, automatic hyperspectral data exploitation by covering the entire end-to-end processing chain from sensor data to end-user products. With the maturing of airborne hyperspectral sensors and the imminent launch of many satellites with hyperspectral imaging capabilities, we are on the threshold of operationally exploiting this data for commercial and military uses. While some researchers focus on individual steps in this chain, such as atmospheric correction, precision geometric processing, or subpixel analysis techniques, the Geobox and Hawkeye tools described here are a compilation and integration of the best available research from one end of the chain to the other. Geobox takes framed image data from a variety of sensors and performs a user-configurable set of geometric and radiometric corrections including geocoding, coregistration, orthorectification, mosaicking, and atmospheric correction. Hawkeye integrates a range of algorithms into a common framework to detect, classify, and map forest changes, urban construction, disturbed soil, and targets of military concern.