Investigation and application of fitting models for centering algorithms in astrometry

To determine the precise positions of stars in CCD frames, various centering algorithms have been proposed for astrometry. The effective point spread function (ePSF) and the Gaussian centering algorithms are two representative centering algorithms. This paper compares in detail and investigates these two centering algorithms in performing data reduction. Specifically, synthetic star images in different conditions (i.e. profiles, fluxes, backgrounds and full width at half maximums) are generated and processed. We find that the difference in precision between the two algorithms is related to the profiles of the star images. Therefore, the precision comparison results using an ideal Gaussian-profile star image cannot be extended to other more specific experimental scenarios. Based on the simulation results, the most appropriate algorithm can be selected according to the image characteristics of observations, and the loss of precision of other algorithms can be estimated. The conclusions are verified using observations captured by the 1-m and 2.4-m telescopes at Yunnan Observatory.