<title>Resolution Classifier</title>

Abstract

A new automated astronomical image classifier is described. The classifier is of the Bayesian type using maximumlikelihood template fitting with Poisson noise. The method's advantages are that there is no need for an explicit galaxy model, it provides a continuous spectrum between totally unresolved objects and obviously diffuse resolved galaxies, and it mn assign a probability to the classification. The continuous nature of the classifier allows identification of intermediate types such as stellar objects with faint nebulosity and galaxies with bright unresolved nuclei. The ability to as sign a probability to each classification allows a determination of when the noise, plate quality, and scale of the images no longer gives a sensible division of stars and galaxies. Also the probability allows the weighting of objects in statistical studies relying on this separation, The method is applied to the catalog of 4 -meter prime focus plates automatically reduced by the FOCAS system It is compared with the hyperserfaoe clustering classifier of Jarvis and Tyson. P(C, IN,) = P(N IC,) P(C5)/P(A),where P(NL IC5) is the probability that the measurements N; are obtained given an object of class Ci , P(Ci) is the a priori probability that the object is of dass Ci regardless of the measurements, and P(NN) is the probability of getting measurements N. For an unbiased classifier all the a priori probabilities are equal and the dass for which P(Ci INL) is maximum is also the one for which P(N; ICJ) is maximum The reason for this little manipulation is that the latter probability can be computed from the noise characteristics of Abstract A new automated astronomical image dassiner is described. The classifier is of the BayesLan type using rnaxirnumlikelihood template fitting with Poisson noise. The method's advantages are that there is no need for an explicit galaxy model, it provides a continuous spectrum between totally unresolved objects and obviously diffuse resolved galaxies, and it can assign a probability to the dassification. The continuous nature of the classifier allows identification of intermediate types such as stellar objects with faint nebulosity and galaxies with bright unresolved nuclei. The ability to assign a probability to each dassLfication allows a determination of when the noise, plate quality, and scale of the images no longer gives a sensible division of stars and galaxies. Also the probability allows the weighting of objects in statistical studies relying on this separation The method is applied to the catalog of 4-meter prime focus plates automatically reduced by the FOCAS system It is compared with the hypersurface clustering dasstfier of Jarvis and Tyson.

A wide field survey at the Northern Ecliptic Pole

“…Cosmic ray hits usually have higher values of conc, since their shape is not determined by the PSF and a cosmic ray event usually affects only a few pixels. Conc is easily derived from the FOCAS-parameters Lc (core luminosity), ispht (isophotal brightness) and ssbr (sky-noise) together with the object brightness mag (the nomenclature of the FOCAS-parameters follows Valdes 1982).…”

“…It is based on the FOCAS-resolution classifier (Valdes 1982). This classifier fits a series of templates, which are basically derived by scaling the width of the image PSF to each object.…”

The Environments of Radio Galaxies

“…Lasker et al 1990). The FOCAS software (Valdes 1982) was used to detect and classify all objects around the target. We calculate No.5, the number of galaxies within 0.5 Mpc of the target with m < 7713 + 2 where 7713 is the 3rd brightest (cluster) galaxy, corrected for background (Bahcall 1981).…”

Deblending and classifying astronomical sources with Mask R-CNN deep learning

“…The task of star/galaxy classification is a long-standing problem in astronomy, dating back to the likes of Messier (1781). Until recently, a technique known as morphological separation (Sebok 1979;Valdes 1982) was the popular choice, which involved a simple assumption: galaxies are resolved sources, and stars point sources. Sebok (1979) and Valdes (1982) pioneered a Bayesian approach focusing on classifying objects by maximizing the probability of object class models matching the observed pixel intensity distributions.…”

“…Until recently, a technique known as morphological separation (Sebok 1979;Valdes 1982) was the popular choice, which involved a simple assumption: galaxies are resolved sources, and stars point sources. Sebok (1979) and Valdes (1982) pioneered a Bayesian approach focusing on classifying objects by maximizing the probability of object class models matching the observed pixel intensity distributions. In contrast, Jarvis & Tyson (1981) use a parametric E-mail: colinjb2@illinois.edu † E-mail: paleo2@illinois.edu method, where clustering of data points of measured pixel intensity distribution determines the classification.…”