Evaluation of centroiding algorithms for an autonomous star tracker

TL;DR

This paper compares six centroiding algorithms for an autonomous star tracker, evaluating their accuracy and computational efficiency through simulations that include noise and Gaussian point spread functions.

Contribution

It provides a comparative analysis of lightweight centroiding algorithms against a shape fitting method for star trackers, considering practical noise and PSF effects.

Findings

Lightweight algorithms show competitive accuracy with shape fitting.

Shape fitting algorithm is more computationally intensive.

Results inform optimal algorithm choice for star tracker design.

Abstract

This work presents numerical results of a computer simulation performed with six centroiding algorithms targeting a star tracker in development at INPE, including readout noise and considering a Gaussian point spread function. Five of the tested algorithms are light-weight centroiding algorithms with low computational costs. These were compared to a shape fitting algorithm based on the lsqnonlin function available in Matlab and GNU Octave. The algorithms studied here are also applicable for astrometry and adaptive optics.