J/A+A/710/A366 KiDS grav. lenses obscured by foreground light (Liu+, 2026)
Identification of gravitational lenses obscured by foreground light in the KiDS
dataset using U-Nets and ResNets.
Liu S., Li R., Jia J., Li H., Chen L., Cao X., He Z., Busillo V.,
Napolitano N.N., Tortora C., Zhong F., Su H., Feng H., Dong Y., Li R.,
Gao L.
<Astron. Astrophys. 710, A366 (2026)>
=2026A&A...710A.366L 2026A&A...710A.366L (SIMBAD/NED BibCode)
ADC_Keywords: Gravitational lensing ; Galaxies ; Optical
Keywords: gravitational lensing: strong - methods: data analysis
Abstract:
Many lensing images are often obscured by foreground light from the
central galaxies, making them challenging to detect.
To address the limitations of previous lens search efforts,
particularly for samples with smaller RE or faint lensed images, we
developed a composite CNN framework that utilizes both U-Net and
ResNet architectures for feature extraction and classification.
We propose a hybrid search method that combines U-Net and ResNet
architectures to enhance the detection of the foreground
galaxy-obscured lenses. Our approach consists of two main stages:
first, the U-Net model separates the foreground galaxy light from
potential lensing signals, creating residual images that highlight the
lensing features. Next, the ResNet module performs binary
classification on these residual images to determine the presence of
lensing signals.
We evaluated the hybrid search method with real observational data to
demonstrate its effectiveness, achieving a recall of 71.5% and a 4.5%
false positive rate at a confidence threshold of 0.6. By applying this
method to over 638398 galaxy samples from the Kilo-Degree Survey Data
Release 4 and conducting thorough inspections, we identified 88 Class
A, 322 Class B, and 1758 Class C candidates.
This hybrid approach significantly enhances the completeness of
existing strong gravitational lensing searches and shows great
potential for improving future astronomical surveys.
Description:
The catalog presents the whole candidates position parameters, the
classification of the visual inspection, the match results of the
referee studies in the paper. The catalog includes five columns, which
respectively represents the KiDS tile identifier, right ascension
(J2000.0), declination (J2000.0), the lensing probability given by
network with the range of 0-1, the class given after visual inspection
e.g. A, B, C.
File Summary:
--------------------------------------------------------------------------------
FileName Lrecl Records Explanations
--------------------------------------------------------------------------------
ReadMe 80 . This file
cand.dat 50 2168 Canditates and match catalog
--------------------------------------------------------------------------------
See also:
II/383 : The ESO KiDS-DR5 multi-band source catalog (Wright+ 2024)
Byte-by-byte Description of file: cand.dat
--------------------------------------------------------------------------------
Bytes Format Units Label Explanations
--------------------------------------------------------------------------------
1- 16 A16 --- KIDS-TILE KiDS tile identifier
18- 27 F10.6 deg RAdeg Right ascension (J2000.0)
29- 38 F10.6 deg DEdeg Declination (J2000.0)
40- 48 F9.7 --- Prob [0/1] Detection probability
50 A1 --- Class [ABC] Class
--------------------------------------------------------------------------------
Acknowledgements:
Sizhuang Liu, 2023202020001(at)whu.edu.cn
License: CC-BY-4.0
(End) Patricia Vannier [CDS] 10-May-2026