A new star identification using patterns in the form of Gaussian mixture models

Abstract

This paper presents an innovative star identification algorithm specifically designed for lost-in-space scenarios in satellite missions. The algorithm introduces a unique approach by utilizing singular values to generate Gaussian mixture models as distinctive features for star identification. By employing a matching process involving candidate filtering, similarity comparison, and validation, the algorithm demonstrates remarkable performance in precisely identifying stars while minimizing false identifications even in demanding environmental conditions. Extensive simulations are conducted to evaluate the algorithm’s effectiveness under various scenarios with positional error and the presence of false stars. The algorithm exhibits reliable and consistent performance, making it highly suitable for dynamic scenarios and contributing to the reduction of hardware burden in star sensor systems. Furthermore, the proposed algorithm offers prominent performance across varying field of view sizes, thereby enhancing its practicality and usability in a wide range of missions.