A quantitative analysis of source detection approaches in optical, infrared, and radio astronomical images

Abstract

A variety of software is used to solve the challenging task of detecting astronomical sources in wide field images. Additionally, computer vision methods based on well-known or innovative techniques are arising to face this purpose. In this paper, we review several of the most promising methods that have emerged during the last few years in the field of source detection. We specifically focus on methods that have been designed to deal with images with Gaussian noise distributions. The singularity of this analysis is that the different methods have been applied to a single dataset consisting of optical, infrared, and radio images. Thus, the different approaches are applied on a level playing field, and the results obtained can be used to evaluate and compare the methods in a meaningful, quantitative way. Moreover, we present the most important strengths and weaknesses of the methods for each type of image as well as an extensive discussion where the methods with best performances are highlighted.