Atmospheric effects on testing and calibrating star tracking algorithms

Abstract

Star trackers are usually considered to be the most accurate sensors, able to achieve a sub-arcminute precision. Star tracker algorithms are often tested and validated with simulated space views. Testing the algorithms with real space images is expensive as it requires implementing them on existing in-space star trackers, or launching new satellites. This study shows that those algorithms are usually performing poorly with ground-based sky pictures and that some adaptations are necessary to take into account the atmospheric effects. In order to tackle this issue, this study will start by implementing and testing two published Lost-In-Space algorithms with a simulated sensor to compare their performance against various noise sources. After comparing the space-based generated views with ground-based images, an adaptation for the aforementioned algorithms is proposed. In order to counter the effect of atmospheric extinction, the number of stars visible in the image is increased by modifying the field-of-view of the camera, the exposure time and estimating the experimental inter-star angular distance error. The idea is to match the star density used in the state-of-the-art algorithms in the experimental pictures. The modified algorithms are tested with the experimental images, and the adaptation process is validated with a good success rate.