Exoplanet imaging data challenge: benchmarking the various image processing methods for exoplanet detection

Cantalloube, F.; Gomez-Gonzalez, C. A.; Absil, O.; Cantero, C.; Bacher, Raphael; Bonse, Markus J.; Bottom, Michael; Dahlqvist, Carl-Henrik; Desgrange, C.; Flasseur, Olivier; Fuhrmann, Thomas; Henning, Th.; Jensen-Clem, Rebecca; Kenworthy, Matthew A.; Mawet, Dimitri; Mesa, D.; Meshkat, Tiffany; Mouillet, D.; Müller, André; Nasedkin, E.; Pairet, Benoît; Pierard, Sébastien; Ruffio, Jean-Baptiste; Samland, M.; Stone, Jordan; Droogenbroeck, Marc Van

doi:10.1117/12.2574803

Cited by 8 publications

(5 citation statements)

References 42 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…High contrast imaging with the recently launched JWST 71 and the upcoming Roman Space Telescope coronagraph 72 will in each case demand a new generation of data-driven PSF and detector calibrations beyond those used fruitfully on ground-based data 73 . While we have not attempted this in the present work, ∂Lux can flexibly model aberrations in the multiple intermediate planes of these coronagraphs, and in future work we intend to assess the practical and fundamental limits of multiplane phase retrieval and calibration to improve coronagraphic modelling.…”

Section: Discussionmentioning

confidence: 99%

Differentiable optics with ∂Lux: I—deep calibration of flat field and phase retrieval with automatic differentiation

Desdoigts

Pope

Dennis

et al. 2023

J. Astron. Telesc. Instrum. Syst.

View full text Add to dashboard Cite

.The sensitivity limits of space telescopes are imposed by uncalibrated errors in the point spread function, photon-noise, background light, and detector sensitivity. These are typically calibrated with specialized wavefront sensor hardware and with flat fields obtained on the ground or with calibration sources, but these leave vulnerabilities to residual time-varying or non-common path aberrations and variations in the detector conditions. It is, therefore, desirable to infer these from science data alone, facing the prohibitively high dimensional problems of phase retrieval and pixel-level calibration. We introduce a new Python package for physical optics simulation, ∂ Lux, which uses the machine learning framework Jax to achieve graphics processing unit acceleration and automatic differentiation (autodiff), and apply this to simulating astronomical imaging. In this first of a series of papers, we show that gradient descent enabled by autodiff can be used to simultaneously perform phase retrieval and calibration of detector sensitivity, scaling efficiently to inferring millions of parameters. This new framework enables high dimensional optimization and inference in data analysis and hardware design in astronomy and beyond, which we explore in subsequent papers in this series.

show abstract

Section: Discussionmentioning

confidence: 99%

Differentiable optics with ∂Lux: I—deep calibration of flat field and phase retrieval with automatic differentiation

Desdoigts

Pope

Dennis

et al. 2023

J. Astron. Telesc. Instrum. Syst.

View full text Add to dashboard Cite

show abstract

“…missed detections). Following standard practice in direct imaging, 1,22,25 detections are treated as blobs of one resolution element radius which corresponds to the expected spatial extent of an exoplanetary signature in the data. From TP, FP, and FN, we derive the true positive rate (TPR, i.e., the recall), the false discovery rate (FDR, i.e., the precision), and the F1R score, which is the harmonic mean between TPR and FDR and used as a measure of the precision-recall trade-off:…”

Section: Loss Function and Accuracy Metricsmentioning

confidence: 99%

“…While demonstrating powerful detection capabilities, this algorithm shown to be prone to a high level of false alarms in some cases. 25 Besides, the tuning of hyper-parameters remains a critical point making the operating point difficult to reach.…”

Section: Introductionmentioning

confidence: 99%

“…Intensive testing of PACO, both on public 25 and on private data challenges, shows that PACO is one of the algorithms of choice to process high-contrast observations. Thanks to its unique data-driven modeling of the data accounting for non-stationary spatial correlations, PACO is especially well suited to process observations in which the typical spatial extent of speckles lies in a patch of a few tens of pixels, as it is the case for SPHERE and GPI observations.…”

Section: Introductionmentioning

confidence: 99%

See 1 more Smart Citation

Exoplanet detection in angular differential imaging: combining a statistics-based learning with a deep-based learning for improved detections

Flasseur

Bodrito

Mairal

et al. 2022

Adaptive Optics Systems VIII

View full text Add to dashboard Cite

Direct imaging is an active research topic in astronomy for the detection and the characterization of young substellar objects. The very high contrast between the host star and its companions makes detection particularly challenging. In addition to the use of an extreme adaptive optics system and a coronagraph to strongly attenuate the starlight contamination, dedicated post-processing methods combining several images recorded with the pupil tracking mode of the telescope are needed.In previous works, we have presented the PACO 1-3 algorithm capturing the spatial correlations of the data with a multi-variate Gaussian model whose parameters are estimated in a data-driven fashion at the scale of a patch of a few tens of pixels. PACO is parameter free and delivers reliable detection confidences with an improved sensitivity compared to the standard methods of the field (e.g., cADI, 4 PCA, 5, 6 TLOCI 7 ). However, there is a room for improvement in the detection sensitivity due to the approximate fidelity of the PACO statistical model with respect to the observations. We propose to combine the statistics-based model of PACO with a deep learning approach in a threestep algorithm. First, the data are centered and whitened locally using the PACO framework to improve the stationarity and the contrast in a preprocessing step. Second, a convolutional neural network is trained in a supervised fashion to detect the signature of synthetic sources in the preprocessed science data. The network is trained from scratch with a custom data augmentation strategy allowing to generate a large training set from a single spatio-temporal dataset. Finally, the trained network is applied to the preprocessed observations and delivers a detection map.We apply our method on eleven datasets from the VLT/SPHERE-IRDIS instrument and compare our method with PACO and other baselines of the field (cADI, PCA). Our results show that the proposed method performs on-par with or better than these algorithms, with a contrast improvement up to half a magnitude with respect to PACO.

show abstract

“…A revolution came with extreme adaptive optics facilities (so called "planet finders") combined to optimized coronagraphs, and also because significant efforts were invested in the observational and data reduction strategies [38]. The instruments SPHERE at the VLT [39], GPI at Gemini [40], and SCExAO at Subaru [41] were prolific to resolve structures in protoplanetary and debris disks in scattered light, shedding light on how the matter is distributed in the first stages of planetary formation.…”

Section: Introductionmentioning

confidence: 99%

Observations of circumstellar disks in scattered light with SPHERE at the VLT

Boccaletti¹

2024

Comptes Rendus. Physique

View full text Add to dashboard Cite

The technological developments initiated in the early 21 st century have led to the implementation of "planet finders" instruments on 8-m class telescopes which are in operation since 2014-2015. Such facilities are at the inception of significant progresses in the study of exoplanetary systems by enabling high contrast imaging of the environment of young nearby stars. One of the most productive science in this field is certainly the observations at high angular resolution of circumstellar disks, from the very young gas-rich protoplanetary disks all the way to more elusive dust-rich debris disks. In this paper, we summarize the main results obtained for these two categories of objects focusing on their morphological characteristics, as well as on the measurement of optical properties of the small dust particles to which the near IR spectral range is sensitive to. We mostly provide exemples based on observations performed with the SPHERE instrument at the VLT.Résumé. Les développements technologiques initiés au début du 21 ème siècle ont conduit à la mise en place d'instruments "découvreurs de planètes" sur des télescopes de la classe 8m en fonctionnement depuis 2014-2015. De telles installations sont à l'origine de progrès significatifs dans l'étude des systèmes exoplanétaires en permettant l'imagerie à haut contraste de l'environnement des jeunes étoiles proches. L'une des sciences les plus productives dans ce domaine est certainement l'observation à haute résolution angulaire des disques circumstellaires, depuis les très jeunes disques protoplanétaires riches en gaz jusqu'aux disques de débris plus insaisissables, riches en poussières. Dans cet article, nous résumons les principaux résultats obtenus pour ces deux catégories d'objets en nous concentrant sur leurs caractéristiques morphologiques, ainsi que sur la mesure des propriétés optiques des petites particules de poussière auxquelles la gamme spectrale du proche IR est sensible. Nous présentons surtout des exemples basés sur des observations réalisées avec l'instrument SPHERE au VLT.

show abstract

Exoplanet imaging data challenge: benchmarking the various image processing methods for exoplanet detection

Abstract: imaging data challenge: benchmarking the various image processing methods for exoplanet detection,"

Cited by 8 publications

References 42 publications

Differentiable optics with ∂Lux: I—deep calibration of flat field and phase retrieval with automatic differentiation

Differentiable optics with ∂Lux: I—deep calibration of flat field and phase retrieval with automatic differentiation

Exoplanet detection in angular differential imaging: combining a statistics-based learning with a deep-based learning for improved detections

Observations of circumstellar disks in scattered light with SPHERE at the VLT

Contact Info

Product

Resources

About