A sub-pixel centroid algorithm for star image based on gaussian distribution

Abstract

A novel centroid algorithm is proposed to acquire better centroid performance in a star image. Since photons incident on pixels of stars present a Gaussian point spread function (PSF), so, the centroid locations on the x and y-axis are calculated using the pixel intensity ratio and expand the Gaussian PSF to polynomials with respect to the centroid location. The angular separation error between star pairs is used to verify the proposed algorithm indirectly. The result clearly shows that the centroid accuracy achieves 1/33 of a pixel and is about two times better than the moment method, indicating the new algorithm is effective. © 2011 The Japan Society for Aeronautical and Space Sciences.