A Graph Based Algorithm for Star Recognition

A novel method is proposed to identify a star into the sky with respect to a reference guidance star database. Contrary to the grid based method, our approach relies on the definition of a graph build on the star local distribution. Via the method, each star is located into its own polygon using the Voronoï partition. Geometric features are extracted from these polygons to form a a star signature used to address the guidance star database. Simulations indicate that the novel method removes the limit of the grid algorithm related to stars located in the same cell of the grid. Our graph-based method exhibits also a better robustness to position and intensity noise when compared to the grid approach.