Least Asymmetry Centering Method and Comparisons

Abstract

The interpretation of astronomical photometry, astrometry, and orbit determination data depends on accurately and consistently identifying the center of the target object’s photometric point spread function in the presence of noise. We introduce a new technique, called least asymmetry, which is designed to find the point about which the distribution is most symmetric. This technique, in addition to the commonly used techniques Gaussian fitting and center of light, was tested against synthetic datasets under realistic ranges of noise and photometric gain. With subpixel accuracy, we compare the determined centers to the known centers and evaluate each method against the simulated conditions. We find that in most cases center of light performs the worst, while Gaussian fitting and least asymmetry are alternately better under different circumstances. Using a real point response function with “reasonable signal-to-noise,” we find that least asymmetry provides the most accurate center estimates, and Gaussian centering is the most precise. The least asymmetry routine implemented in the Python Programming Language can be found at https://github.com/natelust/least_asymmetry.