Aperture Photometric Accuracy of Point-spread-function-deconvolved Astronomical Images

arXiv:2611.01670 · astro-ph.IM · Submitted 2026-11-01 · Read on arXiv

Listen

Radio episode about this paper

Transcript

Introduction to the show: ident: Astrophysics. The day in astrophysics: observations, models and what the data forces.

Vera: We're looking at the paper: "Aperture Photometric Accuracy of Point-spread-function-deconvolved Astronomical Images".

Jocelyn: Here's the gist of the paper. This paper introduces an improved Richardson–Lucy (RL) framework utilizing negative backfilling (NB) and a novel filtered image deconvolution and noise backfilling (FiDNoBa) algorithm to enhance aperture photometric accuracy in astronomical images, which is vital for accurate flux measurement across various source types The results confirm that our method reliably handles complex instrumental signatures, such as the extended wings of the STPSF, mitigating the severe ringing artifacts typical of traditional RL deconvolution while ensuring highly stable aperture photometry in practical survey processing Core Methodology The study improves upon the Richardson–Lucy (RL) algorithm by introducing negative noise backfilling (NB) to better preserve flux conservation, which is necessary when dealing with background-subtracted images The authors propose the Filtered Image Deconvolution and (Filtered-out) Noise Backfilling (FiDNoBa) algorithm, which strategically combines a lowpass prefiltering stage—specifically employing a Kalman filter—with the reintegration of filtered-out noise components after RL iterations This approach involves applying a small-kernel low-pass filter to obtain a filtered image and then separating it into negative and nonnegative parts before applying RL deconvolution to the nonnegative part Finally, the algorithm restores the flux lost during filtering by adding back the previously separated filtered-out noise to synthesize the final image Performance Across Source Regimes The efficacy of these methods was extensively evaluated using numerical simulations across three distinct regimes to demonstrate superior photometric accuracy The evaluation involved isolated point sources (Mock-I), close Sérsic galaxies (Mock-II), and crowded stellar fields (Mock-III) Simulations showed that the Kalman + RL + NB approach demonstrated superior performance across all tested conditions, achieving the lowest relative mean deviation (RMD) Specifically, for isolated sources, this combination reduced the RMD to below 1% and maintained high fidelity for blended Sérsic galaxies In crowded stellar fields, the method achieved approximately 5% RMD while significantly increasing source recovery rates compared to traditional point-spread function photometry Algorithm Comparison and Noise Suppression The FiDNoBa algorithm effectively suppresses noise amplification characteristic of RL iterations by combining low-pass filtering with noise backfilling The authors tested five different low-pass filters, including Kalman, mean, median, BM3D, and NLM to assess their impact on the final result In simulations involving crowded fields (Mock-III), the Kalman + RL + NB combination achieved a relative mean deviation of 5.

Title and authors: Vera: So, focusing back on the title "Aperture Photometric Accuracy of Point-spread-function-deconvolved Astronomical Images," what does that really mean for the people using this data?

Jocelyn: It means if you've done the hard work of deconvolution to get a clearer picture, this paper is about making sure the final measurement—the aperture photometry—is as accurate as possible.

Subrahmanyan: For structure formation studies, accurate flux measurements are everything because they directly translate into reliable estimates of mass and density in galaxies.

Vera: Precisely. Traditional methods often struggle when the instrumental effects or the source structure are complex, leading to systematic errors in those aperture measurements that you can't always see right away.

Jocelyn: This paper suggests a specific way to handle that complexity by integrating negative backfilling into the Richardson-Lucy framework, which is what they call it on page three <ref:2611.01670#pg1>.

Subrahmanyan: It sounds like they are addressing a known limitation in iterative deconvolution techniques regarding how they handle the noise structure inherent in real astronomical data.

Vera: The improvements section explains that they add one step to the traditional nonnegative constraint method by separating the zero-mean image into a negative part, which is mostly noise, and a non-negative part.

Jocelyn: Then they apply Richardson-Lucy deconvolution only to that non-negative part, and after each iteration, they add back the previously separated negative noise image.

Subrahmanyan: That process sounds like it’s designed specifically to ensure that the flux conservation of the RL deconvolution is maintained for aperture photometry, which is what they call it on page three <ref:2611.01670#pg1>.

Vera: They also design the number of Richardson-Lucy deconvolution iterations such that the curve of growth, or encircled energy as a function of radius, becomes horizontal or reaches the actual input flux within one sigma of the deconvolution Gaussian kernel.

Jocelyn: That stopping criterion is important because it tells us when to stop iterating so we don't over-deconvolve and introduce artifacts into our measurement.

Subrahmanyan: It sounds like they are setting a practical boundary based on the expected point spread function behavior, which gives a concrete stopping point instead of relying solely on visual inspection.

Vera: They designed a toy mock test, Mock-zero using an isolated point source with an input flux of six thousand counts that was convolved with a pixel spread function having a full width at half maximum of eight pixels <ref:2611.01670#pg3,designed a toy mock test, Mock-0, using an isolated point source>.

Jocelyn: That setup gives us concrete numbers to see exactly how the algorithm behaves before we even look at the more complex simulations.

Subrahmanyan: Having those concrete input parameters allows us to test the fundamental behavior of this new iterative scheme against known cases before applying it to real survey data.

Vera: The paper then shows that this combination performs well across isolated point sources, reducing relative mean deviation to below one percent in their simulation.

Jocelyn: That reduction is significant because it means we can trust the flux measurements for simple sources much more reliably when using this technique.

Subrahmanyan: A one percent deviation is very low for photometric accuracy in a survey context, suggesting that this method offers a stable baseline performance.

The paper's summary: Vera: So, what’s the big picture takeaway from the summary of this paper about its overall implications?

Jocelyn: It boils down to using negative backfilling and the FiDNoBa algorithm to enhance aperture photometric accuracy in astronomical images, which is vital for accurate flux measurement across various source types.

Subrahmanyan: So, it's a method designed to handle complex instrumental signatures and prevent the severe ringing artifacts typical of traditional RL deconvolution while ensuring highly stable aperture photometry in practical survey processing.

Vera: They are specifically targeting those extended wings of things like the STPSF, which can cause problems with standard RL algorithms.

Jocelyn: The summary emphasizes that this approach reliably handles those complex instrumental signatures, giving researchers confidence that their flux measurements will be stable regardless of the source type being observed.

Subrahmanyan: This suggests that as we process more complex astronomical surveys with more varied source morphologies, this technique provides a robust tool for maintaining consistency in flux extraction.

Vera: It’s about taking the general idea of deconvolution and adding specific steps to preserve flux conservation when you are working with background-subtracted data.

Jocelyn: They are essentially providing a refined toolkit for photometric analysis that addresses known weaknesses in standard algorithms by introducing structured noise reintegration.

Subrahmanyan: This kind of refinement is what we need when we try to connect observational results to cosmological models, because noisy flux measurements introduce uncertainty into those larger calculations.

Vera: It’s about bridging the gap between the raw deconvolution output and a reliable physical measurement you can actually use for science.

The paper's improvements: Jocelyn: Now let's look at the specific improvements they propose to this framework, because that's where the actual engineering happens.

Vera: The main improvement is negative backfilling implementation, which allows Richardson-Lucy deconvolution to better preserve flux conservation by separating negative noise.

Subrahmanyan: Separating the zero-mean image into a negative part, which is mostly noise, and a non-negative part allows you to apply Richardson-Lucy deconvolution only to the non-negative part.

Jocelyn: After each iteration, they add back that previously separated negative noise image back into the result. This ensures the flux conservation of the RL deconvolved image for aperture photometry.

Vera: That whole sequence sounds like a deliberate design choice to ensure that you get flux conservation when you're doing aperture photometry, which is what they call it on page three <ref:2611.01670#pg1>.

Subrahmanyan: The FiDNoBa algorithm is defined as combining lowpass prefiltering with RL deconvolution and noise backfilling to effectively mitigate the noise amplification characteristic of RL iterations while preserving fine spatial details and photometric integrity.

Jocelyn: They tested five different lowpass filters, including Kalman, mean, median, BMthree dee and NLM to see which one had the best impact on the final result.

Vera: And in simulations involving crowded fields—Mock-III—the Kalman plus Richardson-Lucy plus negative backfilling combination achieved a relative mean deviation of five percent there.

Subrahmanyan: That five percent error figure for Mock-III is a concrete number that tells us how much stability we can expect when dealing with dense sources.

Jocelyn: This also shows that the combination is effective even when the source morphology gets very complicated, which is exactly what we want in real sky observations.

Vera: They also mentioned that this process produces images exhibiting a high plateau in their curve of growth near the true flux value, which is beneficial for robust aperture photometry.

Subrahmanyan: A high plateau in the curve of growth near the true flux value is a strong indicator that the final measurement is not overly sensitive to small fluctuations around the peak flux.

Conclusion: Jocelyn: So, to wrap up, what’s our final assessment of these results concerning this paper on "Aperture Photometric Accuracy of Point-spread-function-deconvolved Astronomical Images"?

Vera: The main implication is that this method provides a way to reliably handle complex instrumental signatures and noise amplification in deconvolution, leading to highly stable aperture photometry.

Subrahmanyan: If you can get reliable flux measurements across various source types using this technique, it’s a huge help when you start trying to constrain cosmological models with those observations.

Jocelyn: They showed that the Kalman plus Richardson-Lucy plus negative backfilling combination achieves the lowest relative mean deviation across nearly all test conditions.

Vera: For isolated point sources, they saw a reduction in relative mean deviation to below one percent, which is a very solid result for precision photometry.

Subrahmanyan: And even in crowded areas like Mock-III, where traditional PSF-fitting photometry misses sources, this method helped recover fifty-eight sources where others missed twenty-four faint or blended ones.

Jocelyn: It seems like the FiDNoBa algorithm is a practical way to clean up the output of deconvolution for real survey processing.

Vera: The paper gives us a concrete framework, and we can start testing these concepts on our actual survey data right away to see how much better our flux measurements become.

Subrahmanyan: We’re looking forward to seeing how this approach integrates with larger pipelines, because getting reliable fluxes is the first step toward making stronger statements about the universe's structure.

Shen Zhang, Lei Wang, Yachong Diao, Zhaojun Yan, Xiyan Peng, Yao Liu, Huanyuan Shan, Guanghuan Wang, Fengshan Liu, Chengliang Wei, Lin Nie, Ángkai Ding, Ánya Zheng

National Astronomical Observatories, Chinese Academy of Sciences · School of Astronomy and Space Science, University of Chinese Academy of Sciences · Purple Mountain Observatory, Chinese Academy of Sciences · Beijing City University · Shanghai Astronomical Observatory · School of Physical Science and Technology, Southwest Jiaotong University

astro-ph.IM

Submitted: 2026-11-01

Updated: 2026-11-01

Comments: 20 pages, 13 figures

Journal ref: The Astronomical Journal, 172:67 (18pp), 2026 July

DOI: 10.3847/1538-3881/ae7718

License: http://creativecommons.org/licenses/by/4.0/

Importance score: 90/100

The gist: This paper introduces an improved Richardson–Lucy (RL) framework utilizing negative backfilling (NB) and a novel filtered image deconvolution and noise backfilling (FiDNoBa) algorithm to enhance…

Key concepts

Richardson-Lucy (RL) Algorithm
This is a mathematical technique used to deconvolve images, essentially reversing the blurring caused by the telescope's point spread function. It works iteratively, refining an estimate of the true object by comparing its blurred version to the actual observed image.
Negative Backfilling (NB)
This technique is added to RL to ensure that flux is conserved during deconvolution. When dealing with background-subtracted images, NB separates negative noise components from positive signal, allowing the RL process to accurately estimate the true flux without introducing artifacts.
Filtered Image Deconvolution and Noise Backfilling (FiDNoBa)
This novel algorithm combines a low-pass filter (like a Kalman filter) with RL deconvolution. It first smooths the image to reduce noise amplification, then applies RL only to the positive parts, and finally restores lost flux by reintegrating the filtered-out noise components.

Terminology

Summary

This paper introduces an improved Richardson–Lucy (RL) framework utilizing negative backfilling (NB) and a novel filtered image deconvolution and noise backfilling (FiDNoBa) algorithm to enhance aperture photometric accuracy in astronomical images, which is vital for accurate flux measurement across various source types The results confirm that our method reliably handles complex instrumental signatures, such as the extended wings of the STPSF, mitigating the severe ringing artifacts typical of traditional RL deconvolution while ensuring highly stable aperture photometry in practical survey processing

Core Methodology

The study improves upon the Richardson–Lucy (RL) algorithm by introducing negative noise backfilling (NB) to better preserve flux conservation, which is necessary when dealing with background-subtracted images The authors propose the Filtered Image Deconvolution and (Filtered-out) Noise Backfilling (FiDNoBa) algorithm, which strategically combines a lowpass prefiltering stage—specifically employing a Kalman filter—with the reintegration of filtered-out noise components after RL iterations This approach involves applying a small-kernel low-pass filter to obtain a filtered image and then separating it into negative and nonnegative parts before applying RL deconvolution to the nonnegative part Finally, the algorithm restores the flux lost during filtering by adding back the previously separated filtered-out noise to synthesize the final image

Performance Across Source Regimes

The efficacy of these methods was extensively evaluated using numerical simulations across three distinct regimes to demonstrate superior photometric accuracy The evaluation involved isolated point sources (Mock-I), close Sérsic galaxies (Mock-II), and crowded stellar fields (Mock-III) Simulations showed that the Kalman + RL + NB approach demonstrated superior performance across all tested conditions, achieving the lowest relative mean deviation (RMD) Specifically, for isolated sources, this combination reduced the RMD to below 1% and maintained high fidelity for blended Sérsic galaxies In crowded stellar fields, the method achieved approximately 5% RMD while significantly increasing source recovery rates compared to traditional point-spread function photometry

Algorithm Comparison and Noise Suppression

The FiDNoBa algorithm effectively suppresses noise amplification characteristic of RL iterations by combining low-pass filtering with noise backfilling The authors tested five different low-pass filters, including Kalman, mean, median, BM3D, and NLM to assess their impact on the final result In simulations involving crowded fields (Mock-III), the Kalman + RL + NB combination achieved a relative mean deviation of 5.

Green — the paper says it, and the page is printed. Red — no passage carries it, so treat that sentence as unverified. The same check runs on your own text: check a summary against its source, $5.

Improvements for AI systems

  1. Negative backfilling implementation allows RL deconvolution to better preserve flux conservation by separating negative noise, ensuring the flux conservation of the RL-deconvolved image for aperture photometry.

  2. Filtered Image Deconvolution and Noise Backfilling (FiDNoBa) algorithm combines low-pass filtering with RL deconvolution and noise backfilling to effectively mitigat[e] the noise amplification characteristic of RL iterations while preserv[ing] fine spatial details and photometric integrity.

  3. The Kalman+RL+NB combination achieves the lowest RMD across nearly all test conditions, achieving a reduction in relative mean deviation (RMD) by a factor of approximately three for isolated point sources.

  4. The system can perform enhanced source detection in crowded areas, recovering 58 sources in Mock-III where traditional PSF-fitting photometry missed 24 faint or blended ones.

  5. The deconvolution process produces images exhibiting a high plateau in their curve of growth near the true flux value, which is beneficial for robust aperture photometry.

Abstract

This paper introduces an improved Richardson--Lucy (RL) framework utilizing negative backfilling (NB) and the novel filtered image deconvolution and noise backfilling (FiDNoBa) algorithm. By integrating a low-pass prefiltering stage---specifically employing a Kalman filter---with the reintegration of filtered-out noise components after RL iterations, we effectively suppress noise artifacts while maintaining rigorous flux integrity. We extensively evaluate our Kalman + RL + NB approach using both numerical simulations and real observational data. Simulations across three regimes demonstrate superior photometric accuracy, reducing the relative mean deviation (RMD) to below 1% for isolated sources, maintaining high fidelity (RMD about 0.37%) for blended Sérsic galaxies, and achieving about 5% RMD in crowded stellar fields while significantly increasing source recovery rates compared to traditional point-spread function photometry. Crucially, we validate the practical efficacy of FiDNoBa on real space-based (JWST/NIRCam) and ground-based (Very Large Telescope Survey Telescope OmegaCAM) images. The results confirm that our method reliably handles complex instrumental signatures, such as the extended wings of the STPSF, mitigating the severe ringing artifacts typical of traditional RL deconvolution while ensuring highly stable aperture photometry in practical survey processing.

Related papers