A high-accuracy centroid estimation algorithm for star trackers based on micro-pace matching and filtering

Abstract

The attitude determination accuracy of the star tracker is affected by many factors, among them, the centroiding accuracy of the imaging star point is one of the key factors. In this paper, a high-accuracy centroid estimation algorithm for the star tracker based on micro-pace matching and filtering is proposed. The star points are considered to be discretized sampling of the effective point spread function, in order to restore the it, it is needed to establish reference frames by micro-pace movement. Then, reference frames are interpolated, and accurate reconstruction of the star point energy distribution is obtained. In addition, Point Spread Function(PSF) correlation method is adopted to match a new star point and the reference frame to obtain the centroid displacement. Furthermore, the centroid of the new star point can be calculated. At the same time, the centroid estimation is processed by real-time Extended Kalman filter, which to some extent reduces the influence of noise, and furthermore improves the star point centroid estimation accuracy. The experiment results demonstrate that the proposed approach effectively eliminate the S-curve error of traditional centroid algorithm. The star point estimation accuracy of the proposed approach is approximate 0.0086 pixel, which is a reliable and high-accuracy result for star tracker.