A comparison of feature detection in low atmospheric pressure using thermal and visible images

Abstract

Due to the proliferation of orbital debris, it is now more crucial for spacecraft to be equipped with autonomous obstacle avoidance capabilities. In this paper, we propose a novel optical navigation technique to detect nearby debris efficiently and accurately in real-time by combining feature detection results from thermal and visible images. Furthermore, to demonstrate the robustness of the proposed approach, we compare our results using only thermal or visible imagery at both sea level and at high altitude to show the accuracy gains that can be achieved. The performance of the object detection algorithm was evaluated by comparing the trade-off between object detection accuracy and overall detection runtime across different atmospheric pressures using the Oriented Fast and Rotated Brief (ORB) feature detector.