Improved atmospheric compensation of hyperspectral imagery using LIDAR

Atmospheric compensation of hyperspectral data is important to a number of critical applications such as spectral unmixing and material identification. Current methods perform well at mitigating effects due to sunlight, skylight, and upwelled radiance, but these methods have no information about the geometric properties of the area being imaged. These geometric properties such as line-of-sight to the sun and sky are typically unknown to the standard atmospheric compensation techniques. This research proposes using Light Detection and Ranging (LIDAR) systems to provide a three-dimensional model of the area being imaged. This geometric model is combined with methods such as ELM to provide an improved atmospheric compensation technique that cannot be done with ELM alone. Results demonstrate the improved estimation of reflectance signatures and their effect on subpixel detection applications.