arxiv: 2605.13735 · v2 · submitted 2026-05-13 · 🌌 astro-ph.GA · astro-ph.CO· astro-ph.IM

Recognition: no theorem link

A New PSF Deconvolution Algorithm: Simultaneous Spatial Resolution Enhancement and Point Source Removal for Morphological Analysis of AGN Host Galaxies

Ren Kawase , Takatoshi Shibuya , Kazunori Matsuda

Authors on Pith no claims yet

Pith reviewed 2026-05-15 02:49 UTC · model grok-4.3

classification 🌌 astro-ph.GA astro-ph.COastro-ph.IM

keywords AGN host galaxiesPSF deconvolutionimage reconstructiongalaxy morphologyactive galactic nucleiSubaru Telescopespatial resolution

0 comments

The pith

A deconvolution algorithm enhances the resolution of AGN host galaxy images to Hubble levels while cleanly removing the central point source.

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

The paper introduces a PSF deconvolution method that reconstructs galaxy images by splitting them into an extended smooth host-galaxy component and a sparse point-source AGN component. Three constraints guide the reconstruction: spatial smoothness for the galaxy, sparsity for the point source, and a balance condition from their pixel-wise product to avoid subtraction errors. This approach is tested on artificial and real Subaru Telescope images of AGN at redshifts up to 1, showing resolution gains comparable to space-based observations. If successful, it opens the door to large-scale morphological studies of distant active galaxies using upcoming wide-field surveys.

Core claim

By decomposing an observed image into I_sm for the extended host galaxy and I_sp for the point-source AGN, and enforcing smoothness on I_sm, sparsity on I_sp, and a point-source balance constraint via the product I_sm × I_sp, the algorithm reconstructs images with improved spatial resolution and removed central sources, achieving levels comparable to Hubble Space Telescope data for z ~ 0-1 AGNs observed with Hyper Suprime-Cam.

What carries the argument

The decomposition of the observed image into smooth extended component I_sm and sparse point-source component I_sp, governed by smoothness, sparsity, and point-source balance constraints based on their pixel-wise product.

If this is right

Host-galaxy images from ground-based telescopes reach spatial resolutions matching those from the Hubble Space Telescope.
Bright central AGN point sources are removed without over- or under-subtraction in the host galaxy.
The method enables statistical morphological analyses of distant AGN hosts using data from wide-field surveys like those planned for the Vera C. Rubin Observatory, Euclid, and Roman Space Telescope.

Where Pith is reading between the lines

These are editorial extensions of the paper, not claims the author makes directly.

Similar decomposition techniques could apply to other astronomical images where a bright point source obscures extended structures, such as in exoplanet imaging or crowded stellar fields.
Integration with machine learning could further automate the parameter tuning for the constraints across large datasets.
Testing on higher-redshift AGNs or different filter bands might reveal limitations in the sparsity assumption for more complex nuclear structures.

Load-bearing premise

The point-source balance constraint from the pixel-wise product of the smooth and sparse images removes the central source accurately without introducing artifacts or residuals in real astronomical data.

What would settle it

Applying the algorithm to a set of AGN images with known Hubble Space Telescope counterparts and finding that the recovered host galaxy morphology or central flux residuals differ significantly from the space-based reference.

Figures

Figures reproduced from arXiv: 2605.13735 by Kazunori Matsuda, Ren Kawase, Takatoshi Shibuya.

**Figure 1.** Figure 1: Schematic overview of our method. Blue regions indicate a spatially extended component (e.g., an AGN host galaxy), while yellow regions in the images represent a point source (e.g., an AGN). (Upper panel) Conventional method: PSF deconvolution is applied to the blurred observed image y (left) to obtain the intrinsic image x (right). (Lower panel) Our method: The intrinsic image x is decomposed into two … view at source ↗

**Figure 2.** Figure 2: Dependence of the reconstructed images for an AGN host galaxy on the hyperparameters. The upper-left panels show the reconstructed images in the αsm and αsp space. The red diagonal axis corresponds to the hyperparameter αbl for the point-source balance constraint. See section 2.2 for details. In each set, the left and right images indicate the extended and point-source components, respectively. P(Ism, Isp … view at source ↗

**Figure 3.** Figure 3: Hellinger distance at each iteration for the three targets: the artificial AGN (the blue line), J0959 (the orange dashed line), and J1000 (the green dot-dashed line). See section 5 for details regarding the definition and calculation of the Hellinger distance. 4.2 Real AGN We describe the data acquisition procedure and the properties of the real AGNs. We analyze two real AGNs in the COSMOS field: COSMOS J… view at source ↗

**Figure 4.** Figure 4: Comparison of the original HSC images, the reconstructed images, and the ground-truth Hubble images. The top row shows the artificial AGN, while the middle and bottom rows show the real AGNs, J0959 (z = 0.34) and J1000 (z = 1.38), respectively. In each row from left to right, the panels display: (a) the HSC image, (b) the smooth component Ism, (c) the sparse component Isp, (d) the combined reconstructed im… view at source ↗

**Figure 5.** Figure 5: Comparison of the reconstructed host galaxy (left) and the Hubble PSF-convolved host-galaxy image (right) for the artificial AGN. 6 Results & Discussion This section presents the results of applying the deconvolution method to both artificial and real AGNs. As shown below, qualitative and quantitative evaluations demonstrate that the spatial resolution of HSC is improved to a level comparable to that of … view at source ↗

**Figure 6.** Figure 6: Comparison of the flux profiles. The left panels show the flux profiles along the x-axis passing through the image centers, while the right panels show the profiles passing through the spiral arms. Note that the spiral arm profile is not defined for the artificial AGN because the host galaxy is constructed using a simple Sérsic profile. The red solid line represents the results obtained with the point-sour… view at source ↗

**Figure 7.** Figure 7: Comparison of the reconstructed images obtained from the original pixel-size and the linearly interpolated input images for the real AGNs J0959 (z = 0.34) and J1000 (z = 1.38). The top and bottom rows for each object show the results for the original-size and the interpolated images, respectively. In each row, the panels display the smooth component Ism, the sparse component Isp, and the sum of the smooth … view at source ↗

**Figure 8.** Figure 8: The grid-search results for the artificial AGN. The top, middle, and bottom panels display the smooth component Ism, the sparse component Isp, and the combined reconstructed image Ism + Isp, respectively. Within each αbl block, the columns from left to right correspond to increasing αsp, while the rows from top to bottom correspond to increasing αsm. The numbers shown in the figure indicate the values of (… view at source ↗

**Figure 9.** Figure 9: Same as figure 8, but for J0959 [PITH_FULL_IMAGE:figures/full_fig_p013_9.png] view at source ↗

**Figure 10.** Figure 10: Same as figure 8, but for J1000 [PITH_FULL_IMAGE:figures/full_fig_p014_10.png] view at source ↗

read the original abstract

We propose a new point-spread function (PSF) deconvolution algorithm for images of galaxies hosting an active galactic nucleus (AGN), designed to simultaneously enhance the spatial resolution of the host galaxy and remove the bright central point source. In this algorithm, an intrinsic image is reconstructed by decomposing an observed image into two components: an image $I_{\rm sm}$ of an extended component (i.e., a host galaxy) and an image $I_{\rm sp}$ of a point-source component (i.e., an AGN). During image reconstruction, three constraints are imposed: (1) a smooth constraint on the image $I_{\rm sm}$ , which spatially smooths the host-galaxy structures; (2) a sparse constraint on the image $I_{\rm sp}$ , which localizes the point source to a small number of pixels; and (3) a new constraint, the point-source balance constraint, based on the pixel-wise product $I_{\rm sm} \times I_{\rm sp}$ , which removes the point source from the host galaxy without over- or under-subtraction. As a test, we apply this algorithm to images of artificial and $z \sim 0-1$ real AGNs observed with Hyper Suprime-Cam on the Subaru Telescope. We find that the spatial resolution of the host-galaxy images is improved to a level comparable to that of images from the Hubble Space Telescope and that the bright central point sources are removed. This algorithm is expected to enable statistical morphological studies of distant AGN host galaxies when applied to wide-field survey data from the Vera C. Rubin Observatory, the Euclid Space Telescope, and the Roman Space Telescope.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, simulated authors' rebuttal, and a circularity audit. Tearing a paper down is the easy half of reading it; the pith above is the substance, this is the friction.

Desk Editor's Note private letter to a colleague

The new product-based balance constraint is a practical addition for AGN-host decomposition, but the method's handling of realistic central profiles needs tighter validation.

read the letter

The paper's main contribution is a PSF deconvolution technique that decomposes an observed image into a smooth extended host component and a sparse point-source AGN component. It adds a new balance constraint on the pixel-wise product of those two images to the usual smoothness and sparsity terms, aiming to remove the central source without over- or under-subtraction while sharpening the host to HST-like resolution. That specific three-constraint setup, especially the product term, is the genuinely new element here. The authors apply it to both artificial data and real Subaru Hyper Suprime-Cam images of z~0-1 AGNs and report clean point-source removal plus improved host resolution, which lines up with the needs of upcoming wide-field surveys. The approach is straightforward and could be useful for statistical morphology work on large AGN samples from Rubin, Euclid, or Roman. The soft spot is the one flagged in the stress-test note. The balance term assumes the intrinsic host has little central surface brightness on the PSF scale, but many real galaxies at these redshifts have steep bulges or nuclear clusters. In those cases the optimizer could trade off smoothness against the product penalty and shift flux incorrectly, biasing the recovered host morphology. The abstract and tests lean on visual inspection rather than quantitative recovery metrics such as flux fractions or resolution measures, so it is not yet clear how robust the separation is on realistic profiles. This paper is aimed at extragalactic astronomers who need to clean AGN images for morphological analysis in survey data. Readers working on image-processing tools or AGN host studies would get direct value from the method if the validation holds. It deserves a serious referee because the core idea is grounded in standard constraints plus one targeted addition, and the application to big datasets is timely, even if the current evidence leaves room for refinement on edge cases.

Referee Report

2 major / 0 minor

Summary. The manuscript presents a new PSF deconvolution algorithm for AGN host galaxy images that decomposes the observed image into a smooth extended component I_sm for the host galaxy and a sparse point-source component I_sp for the AGN. Three constraints are applied during reconstruction: smoothness on I_sm, sparsity on I_sp, and a point-source balance constraint using the pixel-wise product I_sm × I_sp to prevent over- or under-subtraction of the point source. Tests on artificial images and real z~0-1 AGN data from Subaru's Hyper Suprime-Cam are reported to show that the host galaxy resolution is enhanced to levels comparable to HST images while removing the central point sources. The method is positioned to enable morphological studies with future wide-field survey data from Rubin, Euclid, and Roman telescopes.

Significance. If quantitatively validated, the algorithm would provide a practical tool for enhancing ground-based imaging resolution and removing AGN contamination, enabling large-scale morphological studies of distant AGN hosts that would otherwise require space-based data. This has clear utility for upcoming wide-field surveys.

major comments (2)

[Abstract and Results] Abstract and Results: No quantitative metrics (e.g., recovered host flux fractions, pre/post-deconvolution FWHM values, residual RMS levels, or error estimates) are reported for the artificial or real-data tests, leaving the central claims of HST-comparable resolution and accurate point-source removal only moderately supported.
[Algorithm description] Algorithm description (point-source balance constraint): The I_sm × I_sp product term assumes negligible central surface brightness in the intrinsic host on PSF-core scales. In galaxies with steep bulges or nuclear star clusters (common at z~0-1), the optimizer may trade smoothness against the balance term, causing flux misallocation. Artificial tests use idealized profiles that do not expose this; real-data examples rely on visual inspection rather than quantitative recovery metrics.

Simulated Author's Rebuttal

2 responses · 0 unresolved

We thank the referee for the constructive and detailed comments, which have helped improve the clarity and rigor of our manuscript. We address each major comment below and have revised the manuscript to incorporate quantitative metrics and additional discussion of the algorithm's assumptions.

read point-by-point responses

Referee: [Abstract and Results] Abstract and Results: No quantitative metrics (e.g., recovered host flux fractions, pre/post-deconvolution FWHM values, residual RMS levels, or error estimates) are reported for the artificial or real-data tests, leaving the central claims of HST-comparable resolution and accurate point-source removal only moderately supported.

Authors: We agree that the absence of quantitative metrics weakens the support for our central claims. In the revised manuscript, we have added comprehensive quantitative metrics for both the artificial and real-data tests. These include recovered host flux fractions, pre- and post-deconvolution FWHM values, residual RMS levels, and associated error estimates. The new results confirm that the host-galaxy resolution reaches levels comparable to HST imaging while the point-source component is accurately removed. revision: yes
Referee: [Algorithm description] Algorithm description (point-source balance constraint): The I_sm × I_sp product term assumes negligible central surface brightness in the intrinsic host on PSF-core scales. In galaxies with steep bulges or nuclear star clusters (common at z~0-1), the optimizer may trade smoothness against the balance term, causing flux misallocation. Artificial tests use idealized profiles that do not expose this; real-data examples rely on visual inspection rather than quantitative recovery metrics.

Authors: We acknowledge the validity of this concern regarding the point-source balance constraint. The assumption of negligible central host surface brightness on PSF-core scales may not hold for galaxies with steep bulges or nuclear star clusters. Although the smoothness constraint on I_sm and sparsity on I_sp are intended to limit flux misallocation, we recognize that idealized artificial tests do not fully probe this regime. In the revised manuscript, we have added an explicit discussion of this assumption and its limitations, together with new tests using simulated galaxies that include steep central profiles. We have also replaced visual inspection of real-data results with the quantitative metrics described in the first response. revision: partial

Circularity Check

0 steps flagged

No significant circularity; constraints defined independently

full rationale

The paper presents an explicit image decomposition into I_sm and I_sp with three directly stated constraints (smoothness on I_sm, sparsity on I_sp, and balance via the pixel-wise product I_sm × I_sp). These are introduced as new algorithmic choices rather than derived from or fitted to the target outputs. Validation uses separate artificial and real Subaru data with visual and resolution comparisons to HST, without reducing any prediction to the inputs by construction or relying on self-citation chains for the core claims. The derivation chain remains self-contained.

Axiom & Free-Parameter Ledger

0 free parameters · 1 axioms · 0 invented entities

The approach rests on standard image-decomposition assumptions in astronomy; no free parameters or invented entities are specified in the abstract.

axioms (1)

domain assumption An observed image can be decomposed into a spatially smooth extended component and a sparse point-source component
Core modeling choice stated in the abstract for separating host galaxy from AGN.

pith-pipeline@v0.9.0 · 5626 in / 1175 out tokens · 27701 ms · 2026-05-15T02:49:41.015256+00:00 · methodology

Review history (2 revisions) →

discussion (0)

Reference graph

Works this paper leans on

59 extracted references · 59 canonical work pages · 1 internal anchor

[1]

2022, PASJ, 74, 247

Aihara, H., et al. 2022, PASJ, 74, 247

work page 2022
[2]

2009, SIAM J

Beck, A., & Teboulle, M. 2009, SIAM J. Imaging Sci., 2, 183

work page 2009
[3]

2025, ApJ, 978, 74

Bonaventura, N., et al. 2025, ApJ, 978, 74

work page 2025
[4]

2026, ApJ, 997, 47

Bonaventura, N., et al. 2026, ApJ, 997, 47

work page 2026
[5]

2011, ApJ, 726, 57

Cisternas, M., et al. 2011, ApJ, 726, 57

work page 2011
[6]

2023, Nature, 616, 45

Chen, Y .-C., et al. 2023, Nature, 616, 45

work page 2023
[7]

Conselice, C. J. 2003, ApJS, 147, 1

work page 2003
[8]

2004, CPAM, 57, 1413

Daubechies, I., et al. 2004, CPAM, 57, 1413

work page 2004
[9]

2019, ApJ, 880, 157

Decarli, R., et al. 2019, ApJ, 880, 157

work page 2019
[10]

2024, A&A, 689, A219 Di Matteo, T., et al

Decarli, R., et al. 2024, A&A, 689, A219 Di Matteo, T., et al. 2005, Nature, 433, 604

work page 2024
[11]

2022, ApJL, 939, L28

Ding, X., et al. 2022, ApJL, 939, L28

work page 2022
[12]

2025, ApJ, 993, 91

Ding, X., et al. 2025, ApJ, 993, 91

work page 2025
[13]

L., et al

Donley, J. L., et al. 2018, ApJ, 853, 63

work page 2018
[14]

2017, A&A, 601, A66

Farrens, S., et al. 2017, A&A, 601, A66

work page 2017
[15]

2000, ApJL, 539, L9

Ferrarese, L., & Merritt, D. 2000, ApJL, 539, L9

work page 2000
[16]

M., et al

Gabor, J. M., et al. 2009, ApJ, 691, 705

work page 2009
[17]

K., et al

Gan, F. K., et al. 2021, arXiv e-prints, arXiv:2103.09711

work page arXiv 2021
[18]

2022, MNRAS, 514, 607

Getachew-Woreta, T., et al. 2022, MNRAS, 514, 607

work page 2022
[19]

2015, ApJ, 806, 218

Glikman, E., et al. 2015, ApJ, 806, 218

work page 2015
[20]

D., et al

Goulding, A. D., et al. 2018, PASJ, 70, S37

work page 2018
[21]

2015, ApJ, 800, 39

Guo, Y ., et al. 2015, ApJ, 800, 39

work page 2015
[22]

M., & Best, P

Heckman, T. M., & Best, P . N. 2014, ARA&A, 52, 589

work page 2014
[23]

F., et al

Hopkins, P . F., et al. 2005, ApJ, 630, 705

work page 2005
[24]

F., et al

Hopkins, P . F., et al. 2006, ApJS, 163, 1

work page 2006
[25]

Photometric Super-Resolution for Improving Galaxy Morphological Measurements using Conditional Generative Adversarial Networks

Kahn, S., et al. 2026, arXiv e-prints, arXiv:2604.20195

work page internal anchor Pith review Pith/arXiv arXiv 2026
[26]

S., et al

Kartaltepe, J. S., et al. 2015, ApJS, 221, 11

work page 2015
[27]

D., et al

Kocevski, D. D., et al. 2012, ApJ, 744, 148

work page 2012
[28]

M., et al

Koekemoer, A. M., et al. 2007, ApJS, 172, 196

work page 2007
[29]

Kormendy, J., & Ho, L. C. 2013, ARA&A, 51, 511 La Marca, A., et al. 2026, A&A, 708, A373

work page 2013
[30]

L., et al

Lambrides, E. L., et al. 2021, ApJ, 919, 129

work page 2021
[31]

T., et al

Leist, M. T., et al. 2024, AJ, 167, 96

work page 2024
[32]

M., et al

Lotz, J. M., et al. 2004, AJ, 128, 163

work page 2004
[33]

1998, ApJ, 494, 472

Magain, P ., et al. 1998, ApJ, 494, 472

work page 1998
[34]

1998, AJ, 115, 2285

Magorrian, J., et al. 1998, AJ, 115, 2285

work page 1998
[35]

A., et al

Marshall, M. A., et al. 2025, A&A, 702, A174

work page 2025
[36]

2018, ApJ, 869, 150

Matsuoka, Y ., et al. 2018, ApJ, 869, 150

work page 2018
[37]

2016, ApJ, 830, 156

Mechtley, M., et al. 2016, ApJ, 830, 156

work page 2016
[38]

2024, AJ, 168, 55

Millon, M., et al. 2024, AJ, 168, 55

work page 2024
[39]

2018, PASJ, 70, S1

Miyazaki, S., et al. 2018, PASJ, 70, S1

work page 2018
[40]

2019, PASJ, 71, 24

Morii, M., et al. 2019, PASJ, 71, 24

work page 2019
[41]

2024, PASJ, 76, 272

Morii, M., et al. 2024, PASJ, 76, 272

work page 2024
[42]

Murata, K., & Takeuchi, T. T. 2022, PASJ, 74, 1329

work page 2022
[43]

2025, Nature Astronomy, 9, 1541

Onoue, M., et al. 2025, Nature Astronomy, 9, 1541

work page 2025
[44]

Y ., et al

Peng, C. Y ., et al. 2002, AJ, 124, 266

work page 2002
[45]

Y ., et al

Peng, C. Y ., et al. 2010, AJ, 139, 2097

work page 2010
[46]

C., et al

Puetter, R. C., et al. 2005, ARA&A, 43, 139

work page 2005
[47]

Richardson, W. H. 1972, J. Opt. Soc. Am., 62, 55

work page 1972
[48]

2017, MNRAS, 467, L110 Sérsic, J

Schawinski, K., et al. 2017, MNRAS, 467, L110 Sérsic, J. L. 1963, BAAA, 6, 41

work page 2017
[49]

2016, ApJ, 821, 72

Shibuya, T., et al. 2016, ApJ, 821, 72

work page 2016
[50]

2025, PASJ, 77, 21

Shibuya, T., et al. 2025, PASJ, 77, 21

work page 2025
[51]

D., & Urry, C

Simmons, B. D., & Urry, C. M. 2008, ApJ, 683, 644

work page 2008
[52]

S., et al

Tanaka, T. S., et al. 2025, ApJ, 995, 21

work page 2025
[53]

1996, JRSSB, 58, 267

Tibshirani, R. 1996, JRSSB, 58, 267

work page 1996
[54]

Tikhonov, A. N. 1963, Soviet Math. Dokl., 4, 1035

work page 1963
[55]

2025, ApJ, 994, 265

Vijarnwannaluk, B., et al. 2025, ApJ, 994, 265

work page 2025
[56]

2017, MNRAS, 466, 812

Villforth, C., et al. 2017, MNRAS, 466, 812

work page 2017
[57]

R., et al

Weaver, J. R., et al. 2022, ApJS, 258, 11

work page 2022
[58]

2022, ApJL, 940, L7

Wylezalek, D., et al. 2022, ApJL, 940, L7

work page 2022
[59]

L., et al

Zakamska, N. L., et al. 2019, MNRAS, 489, 497

work page 2019