Abstract

A star-field pattern recognition algorithm has been developed. It is based on matching the stars on a full celestial sphere observed in a CCD-based star tracker and the reference stars, by using the Yale bright star catalog. The correspondence between the observed and the reference star pattern is found by constructing three-star triangles in spherical geometry. Star angular coordinates and stellar magnitudes are matched to build up a celestial region of correspondence. Gaussian errors in the cataloged star coordinates and brightnesses are modeled to test the algorithm against faulty reference data. Random values are selected for the origin and the orientation of the observed star field. The possibility of hardware-introduced stars is modeled by introducing spurious data. The computer simulations and the preliminary experimental results at the Table Mountain Observatory are described.

© 1992 Optical Society of America