Date of Award

4-4-2017

Document Type

Thesis

Degree Name

Master of Science (MS)

Department

Systems Engineering

First Advisor

Yupo Chan

Abstract

The thrust on Nano-satellite technologies is increasing day by day. Research is going on to detect satellites in austere environment. However, the technology of tracking the satellite in both in-formation and out of formation is yet to discover. I propose a novel algorithm to detect satellites in regular orbit and even in different close operations, i.e. docking, inspection and rendezvous. I aim to modify two existing and promising image processing algorithms and apply towards my own satellite detection problem. Various image processing algorithms have incorporated to enable tracking of the Target Flight Unit (T-FU) from the Chase Flight Unit (C-FU). Two experimental set ups have been modeled for tracking the TFU in-orbit and out of orbit. SURF (Speeded up Robust Feature) and RANSAC (Random Sample Consensus) have been tested in a controlled environment to ensure how we can track the satellite in-orbit. Then color thresholding approach and Bayesian Classifier have been tested for in-orbit operation. Finally, MonoSLAM algorithm has been tested for the depth measurement and the quest for the global co-ordinates of the satellite. Results from the SURF and RANSAC algorithms suggest that tracking at a shorter distance of a small object is possible, however, for long distance the technique should be revised as because SURF algorithm is heavily dependent on features. The results show that it colors thresholding algorithm is appropriate in certain lighting condition where for real time application care must be taken as this it is heavily dependent on the environment and the background image. The results from Bayesian classification suggests that to achieve a precise dynamic threshold more thorough experiments needs to be carried out.

Download

Share

COinS