Challenges
Datasets
Workspaces
Papers
Discussions
Leaderboard
Log inSign up
Challenges
Datasets
Workspaces
Papers
Discussions
Leaderboard
Blog
Job Board
Q3AS

© 2026 Aqora Quantum S.A.S.

TermsPrivacyLegal Notice

arXiv:2610.37491

Adaptive Information Mining: Approaching the Quantum Limit of Super-Resolution Imaging

Seungman Choi, Peter Menart, Andrew Schramka, Zubin Jacob

quant-phphysics.opticsSubmitted Oct 27, 2026• Announced Oct 28, 2026• 0 citations
arXivHTMLPDF

Abstract

The "Rayleigh limit" is not an intrinsic property of light, but a consequence of sub-optimal measurement of collected photons by the imaging instrument. To overcome this limitation, we present Adaptive Information Mining, a framework that achieves super-resolution by using sequentially refined measurements for incoming photons. Our approach employs a reconfigurable imaging system that dynamically reshapes the instrument measurement responses to maximize the Fisher information and guide scene estimation toward the fundamental quantum limit, even in deep sub-Rayleigh regimes. We demonstrate that this approach achieves an estimation error 2-5 times closer to the fundamental quantum limit than traditional direct imaging when imaging point sources, without requiring any prior knowledge of their positions and brightnesses. Importantly, we demonstrate that AIM can be extended to fully blind multi-parameter estimation in scenarios involving an unknown number of sources or turbulence-induced wavefront distortions. Our approach addresses the limitations of conventional fixed imaging, offering a practical path toward quantum-limited super-resolution imaging.

AI Summary

The paper studies adaptive receiver-side measurements for estimating positions and brightnesses of incoherent point sources below the Rayleigh scale, without knowing those parameters beforehand. Adaptive Information Mining (AIM) alternates maximum-likelihood scene estimation from accumulated Poisson photon counts with numerical optimization of spatial-light-modulator (SLM) phases. Two SLMs reshape the detected intensity response; measurement design maximizes the Fisher-information determinant, or minimizes a target parameter’s marginalized Cramér–Rao bound. Evidence consists of optical forward-model simulations and Monte Carlo estimation, rather than an experiment. For three sources with brightness ratio $1:5:20$, five measurements sharing 14,000 photons yield position and brightness errors reported as approximately $2$–$3$ times closer to the quantum bound than direct imaging, including separations down to $0.3R$. This factor measures reduction in the gap above the quantum bound, not absolute error reduction. Targeting separation for two sources with brightness ratio $1:15$ gives an approximately fivefold gap reduction while jointly estimating nuisance parameters. These results use 50 trials per condition. A turbulence extension jointly estimates the scene and wavefront. With $D/r_0=8$, $10^6$ photons and 12 measurements, simulations give approximately $0.2$ rad wavefront error and scene errors below diffraction-limited direct-imaging bounds. Turbulence is assumed frozen throughout acquisition; its estimation uses total-variation regularization, while reported scene CRLBs exclude wavefront uncertainty. Real-time operation requires sufficiently fast hardware and computation. The quantum benchmark assumes multiparameter weak commutativity; the paper notes that attaining it may require collective measurements. Unknown-source-count results are deferred to supplementary material.

Little sign of AI-generated textOverclaimsMissing comparisonSmall numerics onlyStrong assumptions

Error found: The advertised numerical improvement is quantified explicitly using gaps between precision bounds, rather than a reported calculation from empirical estimation errors. Unknown-source-count estimation cannot be independently assessed from the supplied text because its procedures and quantitative evidence are deferred to an unavailable supplement. The turbulence demonstration assumes a frozen, regularized wavefront, and its reported scene-only CRLB does not account for the information penalty from jointly estimating the unknown wavefront.

Written by gpt-6.1-sol from the paper's full text. Automated: it can be wrong.

Comments
Main: 13 pages, 8 figures; Supplement: 15 pages, 11 figures

Related papers

Quantum-limited Optical Super-resolution Imaging of Multiple Point-sources

Kwan Kit Lee, Christos N. Gagatsos, Saikat Guha +1 more

We exploit the quantum parameter estimation approach proposed by Personick [1] to develop a Bayesian adaptive sequential measurement scheme to estimate the number, brightness and location of point sources in a constellation. We demonstrate that the proposed measurement scheme outperforms the traditional diffraction-limited imaging in the deep sub-Rayleigh regime.

Imaging and Applied Optics•Jan 1, 2022•0 citations

Quantum-accelerated imaging of N stars

Fanglin Bao, Hyunsoo Choi, Vaneet Aggarwal +1 more

Imaging point sources with low angular separation near or below the Rayleigh criterion are important in astronomy, e.g., in the search for habitable exoplanets near stars. However, the measurement time required to resolve stars in the sub-Rayleigh region via traditional direct imaging is usually prohibitive. Here we propose quantum-accelerated imaging (QAI) to significantly reduce the measurement time using an information-theoretic approach. QAI achieves quantum acceleration by adaptively learning optimal measurements from data to maximize Fisher information per detected photon. Our approach can be implemented experimentally by linear-projection instruments followed by single-photon detectors. We estimate the position, brightness, and the number of unknown stars 10∼100 times faster than direct imaging with the same aperture. QAI is scalable to a large number of incoherent point sources and can find widespread applicability beyond astronomy to high-speed imaging, fluorescence microscopy, and efficient optical read-out of qubits.

arXiv:2104.14600•Optics Letters•Jun 1, 2021•17 citations

Exoplanet Detection Using Adaptive Quantum-Optimal Measurement

Hyunsoo Choi, Hyoung Won Baac, Zubin Jacob +1 more

Detecting terrestrial exoplanets in the habitable zones of nearby stars remains a critical challenge. Such planets can be \(10^8\) to \(10^{10}\) times fainter than their host stars and lie at diffraction-limited angular separations, where starlight strongly obscures the companion signal. Here we present an adaptive quantum measurement method for estimating the number, positions, and brightnesses of mutually incoherent point sources in the sub-Rayleigh, ultra-high-contrast regime, operating at contrasts down to \(10^{-8}\) -- five orders of magnitude beyond previous quantum imaging approaches to exoplanet detection. The method adopts a spatial-mode basis that is updated to maximize the quantum Fisher information per detected photon. Estimation is performed by maximum likelihood in log-brightness coordinates, and the source count is determined by Bayesian-information-criterion (BIC) model selection directly from photon-count statistics, without a tunable detection threshold. For point sources within sub-Rayleigh separations and with brightness ratios spanning eight orders of magnitude, the method reconstructs complete scenes with a mean success rate of \(72.5\%\). Furthermore, it is robust to misalignment, maintaining a \(71.3\%\) success rate under offsets of up to six pixels. These results demonstrate that terrestrial exoplanets can be detected below the Rayleigh limit, a regime previously inaccessible to direct imaging.

arXiv:2607.06931•Jul 8, 2026•0 citations

Passive optical superresolution at the quantum limit

A. I. Lvovsky, Michael R. Grace, Saikat Guha +3 more

For more than a century, the diffraction limit has defined the resolution achievable by passive optical imaging systems. Although some resolution improvement can be gained through classical data processing of the image, it is limited by the noise arising from quantum nature of light. Minimizing the effect of this noise requires quantum treatment of optical imaging. By reformulating imaging as a problem of quantum measurement and estimation, it becomes possible to identify optimal detection strategies that recover spatial information previously thought inaccessible. This review summarizes the theoretical framework that underpins this development, from the formulation of quantum Cramer-Rao bounds and Chernoff bounds to the construction of receivers that attain them, such as those based on spatial-mode demultiplexing. We show how these methods can beat conventional imaging in the classification, localization, and imaging of sub-Rayleigh incoherent sources. We then discuss extensions to multiparameter and partially coherent scenarios, and highlight the unifying connections between estimation and discrimination tasks. Finally, we survey recent experimental demonstrations that approach quantum-limited resolution and outline emerging applications in microscopy, astronomy, and optical sensing.

quant-pharXiv:2605.10767•Optica•May 12, 2026•0 citations

Quantum Multi-Parameter Adaptive Bayesian Estimation and Application to Super-Resolution Imaging

Kwan Kit Lee, Christos N. Gagatsos, Saikat Guha +1 more

In Bayesian estimation theory, the estimator ${\hat θ} = E[θ|l]$ attains the minimum mean squared error (MMSE) for estimating a scalar parameter of interest $θ$ from the observation of $l$ through a noisy channel $P_{l|θ}$, given a prior $P_θ$ on $θ$. In quantum sensing tasks, the user gets $ρ_θ$, the quantum state that encodes $θ$. They choose a measurement, a positive-operator valued measure (POVM) $Π_l$, which induces the channel $P_{l|θ} = {\rm Tr}(ρ_θΠ_l)$ to the measurement outcome $l$, on which the aforesaid classical MMSE estimator is employed. Personick found the optimum POVM $Π_l$ that minimizes the MMSE over all possible measurements, and that MMSE. This result from 1971 is less-widely known than the quantum Fisher information (QFI), which lower bounds the variance of an unbiased estimator over all measurements, when $P_θ$ is unavailable. For multi-parameter estimation, i.e., when $θ$ is a vector, in Fisher quantum estimation theory, the inverse of the QFI matrix provides an operator lower bound to the covariance of an unbiased estimator. However, there has been little work on quantifying quantum limits and measurement designs, for multi-parameter quantum estimation in the {\em Bayesian} setting. In this paper, we build upon Personick's result to construct a Bayesian adaptive measurement scheme for multi-parameter estimation when $N$ copies of $ρ_θ$ are available. We illustrate an application to localizing a cluster of point emitters in a highly sub-Rayleigh angular field-of-view, an important problem in fluorescence microscopy and astronomy. Our algorithm translates to a multi-spatial-mode transformation prior to a photon-detection array, with electro-optic feedback to adapt the mode sorter. We show that this receiver performs far superior to quantum-noise-limited focal-plane direct imaging.

arXiv:2202.09980•Feb 21, 2022•1 citation

Quantum limited superresolution: imaging beyond the Rayleigh diffraction limit

Sudhakar Prasad, Jonathan R. Owens, J. W. Bray +1 more

Conventional superresolution of closely spaced point sources using image intensity data acquired by noiseless single-photon detectors suffers from Rayleigh’s diffraction curse that can only be overcome at photon numbers that scale quartically with the degree of required superresolution. By contrast, acquiring and processing suitably chosen mode projections of the wavefront, which contain dual information of its intensity and phase, can reduce the photon budget to one that is quadratic in the degree of superresolution and achieve quantum-limited performance. Such dramatic improvements enabled by the wavefront projection (WFP) approach have been realized both in computer simulation and laboratory experiments over the last five years. The invited talk will explore the fundamental, theoretically insuperable limits on superresolution imaging by presenting computations of quantum Fisher information and practical realizations of the novel WFP approach that can reach these limits for a variety of point and extended source geometries and operating conditions.

Optical and Quantum Sensing and Precision Metrology II•Mar 3, 2022•1 citation
The "Rayleigh limit" is not an intrinsic property of light, but a consequence of sub-optimal measurement of collected photons by the imaging instrument. To overcome this limitation, we present Adaptive Information Mining, a framework that achieves super-resolution by using sequentially refined measurements for incoming photons. Our approach employs a reconfigurable imaging system that dynamically reshapes the instrument measurement responses to maximize the Fisher information and guide scene estimation toward the fundamental quantum limit, even in deep sub-Rayleigh regimes. We demonstrate that this approach achieves an estimation error 2-5 times closer to the fundamental quantum limit than traditional direct imaging when imaging point sources, without requiring any prior knowledge of their positions and brightnesses. Importantly, we demonstrate that AIM can be extended to fully blind multi-parameter estimation in scenarios involving an unknown number of sources or turbulence-induced wavefront distortions. Our approach addresses the limitations of conventional fixed imaging, offering a practical path toward quantum-limited super-resolution imaging.
The paper studies adaptive receiver-side measurements for estimating positions and brightnesses of incoherent point sources below the Rayleigh scale, without knowing those parameters beforehand. Adaptive Information Mining (AIM) alternates maximum-likelihood scene estimation from accumulated Poisson photon counts with numerical optimization of spatial-light-modulator (SLM) phases. Two SLMs reshape the detected intensity response; measurement design maximizes the Fisher-information determinant, or minimizes a target parameter’s marginalized Cramér–Rao bound.
Evidence consists of optical forward-model simulations and Monte Carlo estimation, rather than an experiment. For three sources with brightness ratio 1:5:201:5:201:5:20, five measurements sharing 14,000 photons yield position and brightness errors reported as approximately 222–333 times closer to the quantum bound than direct imaging, including separations down to 0.3R0.3R0.3R. This factor measures reduction in the gap above the quantum bound, not absolute error reduction. Targeting separation for two sources with brightness ratio 1:151:151:15 gives an approximately fivefold gap reduction while jointly estimating nuisance parameters. These results use 50 trials per condition.
A turbulence extension jointly estimates the scene and wavefront. With D/r0=8D/r_0=8D/r0​=8, 10610^6106 photons and 12 measurements, simulations give approximately 0.20.20.2 rad wavefront error and scene errors below diffraction-limited direct-imaging bounds. Turbulence is assumed frozen throughout acquisition; its estimation uses total-variation regularization, while reported scene CRLBs exclude wavefront uncertainty. Real-time operation requires sufficiently fast hardware and computation. The quantum benchmark assumes multiparameter weak commutativity; the paper notes that attaining it may require collective measurements. Unknown-source-count results are deferred to supplementary material.
Imaging point sources with low angular separation near or below the Rayleigh criterion are important in astronomy, e.g., in the search for habitable exoplanets near stars. However, the measurement time required to resolve stars in the sub-Rayleigh region via traditional direct imaging is usually prohibitive. Here we propose quantum-accelerated imaging (QAI) to significantly reduce the measurement time using an information-theoretic approach. QAI achieves quantum acceleration by adaptively learning optimal measurements from data to maximize Fisher information per detected photon. Our approach can be implemented experimentally by linear-projection instruments followed by single-photon detectors. We estimate the position, brightness, and the number of unknown stars 10∼100 times faster than direct imaging with the same aperture. QAI is scalable to a large number of incoherent point sources and can find widespread applicability beyond astronomy to high-speed imaging, fluorescence microscopy, and efficient optical read-out of qubits.
Detecting terrestrial exoplanets in the habitable zones of nearby stars remains a critical challenge. Such planets can be 10810^8108 to 101010^{10}1010 times fainter than their host stars and lie at diffraction-limited angular separations, where starlight strongly obscures the companion signal. Here we present an adaptive quantum measurement method for estimating the number, positions, and brightnesses of mutually incoherent point sources in the sub-Rayleigh, ultra-high-contrast regime, operating at contrasts down to 10−810^{-8}10−8 -- five orders of magnitude beyond previous quantum imaging approaches to exoplanet detection. The method adopts a spatial-mode basis that is updated to maximize the quantum Fisher information per detected photon. Estimation is performed by maximum likelihood in log-brightness coordinates, and the source count is determined by Bayesian-information-criterion (BIC) model selection directly from photon-count statistics, without a tunable detection threshold. For point sources within sub-Rayleigh separations and with brightness ratios spanning eight orders of magnitude, the method reconstructs complete scenes with a mean success rate of 72.5%72.5\%72.5%. Furthermore, it is robust to misalignment, maintaining a 71.3%71.3\%71.3% success rate under offsets of up to six pixels. These results demonstrate that terrestrial exoplanets can be detected below the Rayleigh limit, a regime previously inaccessible to direct imaging.
For more than a century, the diffraction limit has defined the resolution achievable by passive optical imaging systems. Although some resolution improvement can be gained through classical data processing of the image, it is limited by the noise arising from quantum nature of light. Minimizing the effect of this noise requires quantum treatment of optical imaging. By reformulating imaging as a problem of quantum measurement and estimation, it becomes possible to identify optimal detection strategies that recover spatial information previously thought inaccessible. This review summarizes the theoretical framework that underpins this development, from the formulation of quantum Cramer-Rao bounds and Chernoff bounds to the construction of receivers that attain them, such as those based on spatial-mode demultiplexing. We show how these methods can beat conventional imaging in the classification, localization, and imaging of sub-Rayleigh incoherent sources. We then discuss extensions to multiparameter and partially coherent scenarios, and highlight the unifying connections between estimation and discrimination tasks. Finally, we survey recent experimental demonstrations that approach quantum-limited resolution and outline emerging applications in microscopy, astronomy, and optical sensing.
In Bayesian estimation theory, the estimator θ^=E[θ∣l]{\hat θ} = E[θ|l]θ^=E[θ∣l] attains the minimum mean squared error (MMSE) for estimating a scalar parameter of interest θθθ from the observation of lll through a noisy channel Pl∣θP_{l|θ}Pl∣θ​, given a prior PθP_θPθ​ on θθθ. In quantum sensing tasks, the user gets ρθρ_θρθ​, the quantum state that encodes θθθ. They choose a measurement, a positive-operator valued measure (POVM) ΠlΠ_lΠl​, which induces the channel Pl∣θ=Tr(ρθΠl)P_{l|θ} = {\rm Tr}(ρ_θΠ_l)Pl∣θ​=Tr(ρθ​Πl​) to the measurement outcome lll, on which the aforesaid classical MMSE estimator is employed. Personick found the optimum POVM ΠlΠ_lΠl​ that minimizes the MMSE over all possible measurements, and that MMSE. This result from 1971 is less-widely known than the quantum Fisher information (QFI), which lower bounds the variance of an unbiased estimator over all measurements, when PθP_θPθ​ is unavailable. For multi-parameter estimation, i.e., when θθθ is a vector, in Fisher quantum estimation theory, the inverse of the QFI matrix provides an operator lower bound to the covariance of an unbiased estimator. However, there has been little work on quantifying quantum limits and measurement designs, for multi-parameter quantum estimation in the {\em Bayesian} setting. In this paper, we build upon Personick's result to construct a Bayesian adaptive measurement scheme for multi-parameter estimation when NNN copies of ρθρ_θρθ​ are available. We illustrate an application to localizing a cluster of point emitters in a highly sub-Rayleigh angular field-of-view, an important problem in fluorescence microscopy and astronomy. Our algorithm translates to a multi-spatial-mode transformation prior to a photon-detection array, with electro-optic feedback to adapt the mode sorter. We show that this receiver performs far superior to quantum-noise-limited focal-plane direct imaging.
Conventional superresolution of closely spaced point sources using image intensity data acquired by noiseless single-photon detectors suffers from Rayleigh’s diffraction curse that can only be overcome at photon numbers that scale quartically with the degree of required superresolution. By contrast, acquiring and processing suitably chosen mode projections of the wavefront, which contain dual information of its intensity and phase, can reduce the photon budget to one that is quadratic in the degree of superresolution and achieve quantum-limited performance. Such dramatic improvements enabled by the wavefront projection (WFP) approach have been realized both in computer simulation and laboratory experiments over the last five years. The invited talk will explore the fundamental, theoretically insuperable limits on superresolution imaging by presenting computations of quantum Fisher information and practical realizations of the novel WFP approach that can reach these limits for a variety of point and extended source geometries and operating conditions.