Date of Award
Master of Science
Department of Engineering Physics
Michael Roggemann, PhD
The objective of this research was to determine if measurements from a Sagnac interferometer could provide reliable estimates of satellite material composition. The Sagnac interferometer yields a spatial interferogram that can sampled by a linear detector array. The interferogram is related to the spectrum of the source through a Fourier transform. Here, spectral reflectivities of nine common satellite materials were used to simulate the spectrum on obtains from an ideal Sagnac interferometer in the beam-train of a ground-based telescope whose mission is to view satellites. The signal-to-noise ratio of the spectrum was varied to simulate the effect of range variation between the sensor and the satellite. The simulated spectra consisted of a linear mixture of spectra from two of the nine materials. Three different architectures were developed and their performances compared. One of the three architectures consisted of nine artificial neural networks (ANNs), one for each material, and a linear estimator that estimated the satellite surface area attributable to each material. This method estimates the material composition by using a classifier to identify the materials contributing to the mixture, then eliminating unlikely contributors to the mixture before performing a constrained linear estimate. It is shown that due to high classification errors, the system using solely a linear estimator provides the estimate with the lowest errors.
DTIC Accession Number
Caudill, Eugene L., "Satellite Surface Material Composition from Synthetic Spectra" (1994). Theses and Dissertations. 6436.