A New Celestial Positioning Model Based on Robust Estimation

A large field of vision (FOV) star sensor can be used to image multiple celestial bodies at the same time, and then through center extraction of the star point and star patterns matching to realize navigation and positioning. The FOV based method has gradually become the main way of celestial navigation. However, in the process of center extraction, mistakenly identifying image noise as the star point and encountering some abnormal coordinate errors are inevitable. Meanwhile, in the process of star patterns matching, wrong matches are always encountered too. To solve these problems, this paper introduces the Robust Estimation for the first time, and based on it a celestial positioning model is established. The experiment shows that the model can effectively exclude the impact of noise and wrong patching, limit the abnormal observations, and take full advantage of high-precision observation information, by which positioning accuracy is increased from 4.05 '' to 1.15 ''.