Automated clustering method for point spread function classification

Abstract

The point spread function (PSF) plays a very important part in image post-processing and high-precion astrometry and photometry. It is necessary to analyse the properties of the PSF before we use it to process data. However, in real observations, the PSF is affected by many different factors and the shape of it has inevitable spatial and temporal variations that can be hard to describe. In this paper, we propose a clustering method to evaluate the shape variations of PSFs. We analyse the performance of this method with simulated PSFs under different observation conditions. Then, we process two observational data sets with this method. The PSF clustering results can provide a reference for checking observation conditions and can be used for astrometry based on PSF fitting. In general, our method can reveal the morphologic similarities of different PSFs and can provide a reference for observations. The cluster revealed by our method can provide a reference for the evaluation of observation conditions and for the post-processing of astronomical observation data.