Optimising and comparing source extraction tools using objective segmentation quality criteria

Abstract

With the growth of the scale, depth, and resolution of astronomical imaging surveys, there is an increased need for highly accurate automated detection and extraction of astronomical sources from images. This also means there is a need for objective quality criteria, and automated methods to optimise parameter settings for these software tools. We present a comparison of several tools which have been developed to perform this task: namely SExtractor, ProFound, NoiseChisel, and MTObjects. In particular, we focus on evaluating performance in situations which present challenges for detection -- for example, faint and diffuse galaxies; extended structures, such as streams; and objects close to bright sources. Furthermore, we develop an automated method to optimise the parameters for the above tools. We present four different objective segmentation quality measures, based on precision, recall, and a new measure for the correctly identified area of sources. Bayesian optimisation is used to find optimal parameter settings for each of the four tools on simulated data, for which a ground truth is known. After training, the tools are tested on similar simulated data, to provide a performance baseline. We then qualitatively assess tool performance on real astronomical images from two different surveys. We determine that when area is disregarded, all four tools are capable of broadly similar levels of detection completeness, while only NoiseChisel and MTObjects are capable of locating the faint outskirts of objects. MTObjects produces the highest scores on all tests on all four quality measures, whilst SExtractor obtains the highest speeds. No tool has sufficient speed and accuracy to be well-suited to large-scale automated segmentation in its current form.

Accepted in Astronomy and Astrophysics. Second version 16 Nov 2020 with a change in the acknowledgements

References in corpus (29)

• Astropy: A Community Python Package for Astronomy
• The Astropy Project: Building an inclusive, open-science project and status of the v2.0 core package
• LSST: from Science Drivers to Reference Design and Anticipated Data Products
• The Fourth Data Release of the Sloan Digital Sky Survey
• The Hubble Ultra Deep Field
• Structural Analysis of Molecular Clouds: Dendrograms
• Forty-Seven Milky Way-Sized, Extremely Diffuse Galaxies in the Coma Cluster
• UV Continuum Slope and Dust Obscuration from z~6 to z~2: The Star Formation Rate Density at High Redshift
• GEMS: Galaxy Evolution from Morphologies and SEDs
• z~7-10 Galaxies in the HUDF and GOODS fields, and their UV Luminosity Functions
• Structure and Morphologies of z~7-8 Galaxies from ultra-deep WFC3/IR Imaging of the HUDF
• Ultra diffuse galaxies outside clusters: clues to their formation and evolution
• ProFound: Source Extraction and Application to Modern Survey Data
• The Fornax Deep Survey with VST. I. The extended and diffuse stellar halo of NGC~1399 out to 192 kpc
• The EFIGI catalogue of 4458 nearby galaxies with detailed morphology
• Noise Based Detection and Segmentation of Nebulous Objects
• Spatial distribution of ultra-diffuse galaxies within large-scale structures
• The IAC Stripe 82 Legacy Project: a wide-area survey for faint surface brightness astronomy
• Galactic cirri in deep optical imaging
• SCARLET: Source separation in multi-band images by Constrained Matrix Factorization
• The Fornax Deep Survey (FDS) with VST. VI. Optical properties of the dwarf galaxies in the Fornax cluster
• The Fornax Deep Survey with VST. V. Exploring the faintest regions of the bright early-type galaxies inside the virial radius
• Photometry of high-redshift blended galaxies using deep learning
• Deblending galaxy superpositions with branched generative adversarial networks
• Convergence of Lobatto-type Runge-Kutta methods for partitioned differential-algebraic systems of index 2
• The missing light of the Hubble Ultra Deep Field
• Automated detection of very Low Surface Brightness galaxies in the Virgo Cluster
• Background sky obscuration by cluster galaxies as a source of systematic error for weak lensing
• The Fornax Deep Survey (FDS) with the VST: III. Low Surface Brightness (LSB) dwarfs and Ultra Diffuse Galaxies (UDGs) in the center of the Fornax cluster

Cited by in corpus (15)

• Introducing the LBT Imaging of Galactic Halos and Tidal Structures (LIGHTS) survey. A preview of the low surface brightness Universe to be unveiled by LSST
• Tracing satellite planes in the Sculptor group: I. Discovery of three faint dwarf galaxies around NGC 253
• The edges of galaxies: Tracing the limits of star formation
• Discovery and analysis of low-surface-brightness galaxies in the environment of NGC 1052
• Inferring the properties of the sources of reionization using the morphological spectra of the ionized regions
• Dissecting the RELICS cluster SPT-CLJ0615-5746 through the intracluster light: confirmation of the multiple merging state of the cluster formation
• The PAU survey: Estimating galaxy photometry with deep learning
• Extragalactic Stellar Tidal Streams in the Dark Energy Survey
• The impact of environment on size: Galaxies are 50% smaller in the Fornax Cluster compared to the field
• The AMIGA sample of isolated galaxies. XIV. Disc breaks and interactions through ultra-deep optical imaging
• A comparative study of source-finding techniques in HI emission line cubes using SoFiA, MTObjects, and supervised deep learning
• Revisiting the structure of galactic disks with deep imaging
• Analysis of Galactic cirrus filaments in HSC-SSP high-resolution deep images using artificial neural networks
• The Fornax Deep Survey (FDS) with the VST: III. Low Surface Brightness (LSB) dwarfs and Ultra Diffuse Galaxies (UDGs) in the center of the Fornax cluster
• The Exploration of Local VolumE Satellites (ELVES) Survey: A Nearly Volume-Limited Sample of Nearby Dwarf Satellite Systems