Hyperspectral Scene Simulation in the Ultraviolet through Longwave Infrared

Efficient design of remote spectral imaging sensors, including the optics and data processing algorithms, depends critically on the availability of well-characterized data that span the full operational space of the sensor. As it is not practical to spanthis space with field data alone, high-fidelity scene simulations are needed. To address this need, Spectral Sciences, Inc. proposes to develop a hyperspectral/multispectral scene simulator for the UV/visible through thermal infrared (TIR) (~0.4-13mm)regions that accurately treats atmospheric effects and incorporates realistic spectral and spatial variability (clutter) in the surface materials. In Phase I we successfully demonstrated a novel method to building synthetic imagery with data-based spatialand spectral surface properties, and formulated an efficient Monte Carlo radiation transport method for high-fidelity image simulations covering the full wavelength range of interest. In Phase II we will fully implement these and other capabilities in aprototype software toolkit and perform extensive validation and analysis with a variety of sensors and data sets. Phase II and Phase III products will be developed in collaboration with a team of commercial and academic partners that have considerableexperience in remote sensing, image processing, and technical software development.