Estimation of Satellite Orientation from Space Surveillance Imagery Measured withan Adaptive Optics Telescope

摘要

The use of the matched filter to automatically estimate the pose of a Low EarthOrbiting satellite from imagery taken with an adaptive optics telescope is explored. This work represents the first effort to solve the satellite pose estimation problem while considering the broad range of atmospheric turbulence levels and target visual magnitudes that are encountered in ground based space surveillance operations. Several Algorithms are examined in an effort to determine the performance bounds on the matched filter for this application. Results are given over an extremely wide range of seeing conditions. These results are weighted based on historical data obtained from the Air Force Maui Optical Station (AMOS) to show that, under normal imaging conditions, the matched filter approach proposed in this thesis can be expected to yield correct pose estimations in over eighty percent of the trials considered. Additionally, it is shown that a significant portion of errors are between two poses that are very similar in appearance, such as views of the target about an axis of symmetry. A correlation value is defined such that a value between 0.0 and 1.0 is assigned for each trial. This correlation value is a quantitative measure of the similarity between the predicted pose and the actual pose being imaged, with a value of 1.0 being representative of a correct pose estimation. When weighted according to actual seeing conditions, the algorithm developed in this study provides an average correlation value in excess of 0.98. A dependable algorithm is defined that will function efficiently under a wide range of seeing conditions that are encountered in ground based space surveillance operations.