Implementation and Analysis of 3-D Tomographic Reconstructions from Space-Based Imaging Platforms

Imaging satellites that look nadir face a variety of obstacles. In addition to designing the system for the intense environment that the satellite will be experiencing, there are other factors to consider: reflections and emissions from the ground, from clouds, and from the OH-airglow layer. Depending on the desired object, these nuisance signals can significantly reduce image quality. The ground will have city lights, clouds will reflect light, and every material will have a different reflectance, some up to 60%. Performing a tomographic reconstruction can effectively separate a signal from other emissions and reflections. The Atmospheric Waves Experiment (AWE) is a prime example for use of tomographic reconstruction techniques from an imaging space platform. AWE is designed for studying the OH-airglow layer and atmospheric waves (also called gravity waves) which cause emission changes in the OH-airglow layer. A reconstruction for AWE would separate signals from the OH-airglow layer from reflected light from clouds and the ground. Performing tomographic reconstructions for the Atmospheric Waves Experiment (AWE) and analyzing them is the primary focus of this dissertation. This work covers an implementation of MLEM for use in satellite images pointing nadir. The algorithm is fast enough to be performed in real-time for many applications. This work covers the details of the reconstruction implementation and the challenges it poses and then a detailed study of the image quality of the tomographic reconstructions is presented. Some of the useful tools developed during this study include the construction of a short-wave infrared (SWIR) model of the atmosphere, methods for projecting simulated models through the imaging system, performing tomographic reconstructions of the simulations, and using a Hotelling observer to determine the overall image quality. Tomographic reconstructions are found to be effective in many applications for space imaging. However, the severely limited projection angles do provide constraints on the overall reconstructed resolution