Comparison of LSBGnet to MTO for Finding Low‐Surface‐Brightness Galaxies

Belu, Oana Miruna (2025) Comparison of LSBGnet to MTO for Finding Low‐Surface‐Brightness Galaxies. Bachelor's Thesis, Computing Science.

Preview	Text bCS2025BeluOM.pdf Download (37MB) | Preview
	Text Toestemming.pdf Restricted to Registered users only Download (193kB)

Abstract

Low-surface-brightness galaxies (or LSBGs) present a significant challenge for detection due to their low central surface brightness. Nevertheless, discovering the most effective methods to detect them is crucial for our advancement in understanding our Universe's composition and dynamics. This project presents a thorough comparison between LSBGnet, a neural network that specializes in the automatic LSBG detection and MTO, a Max-Tree-based approach for detecting astronomical sources, the latter one being able to detect all astronomical objects, not just low-surface-brightness galaxies. The performance of both algorithms is evaluated based on their detection rates across key galaxy parameters, including central surface brightness, effective radius, ellipticity, halo-to-disk ratio, and background density. We find that LSBGnet has an overall detection completeness of 97%, compared to 82% for MTO - with the highest gap at the faintest end at which we tested, while MTO recovers the injected galaxies more faithfully. We also observe that both detection pipelines have zero False Positives when feeding them pure-noise images.

Item Type:	Thesis (Bachelor's Thesis)
Supervisor name:	Wilkinson, M.H.F. and Bunte, K.
Degree programme:	Computing Science
Thesis type:	Bachelor's Thesis
Language:	English
Date Deposited:	16 Jul 2025 07:05
Last Modified:	16 Jul 2025 07:05
URI:	https://fse.studenttheses.ub.rug.nl/id/eprint/36308

Actions (login required)

View Item