Post-processing for AO-corrected images Inverse Problems in Imaging - PowerPoint PPT Presentation

Post-processing for AO-corrected images Inverse Problems in Imaging May 23 rd , 2018 Summer school - Lake Como Supervisors Co-supervisors Thierry FUSCO (ONERA/LAM) Benoit NEICHEL (LAM) Laurent MUGNIER (ONERA) Aurélie BONNEFOIS (ONERA) Romain JL. FETICK 24/05/2018 1 24/05/2018

Summary I. Imaging with adaptive optics in astronomy Paris II. Theory on deconvolution and myopic deconvolution Como III. PSF parametrization IV. Tests on VLT SPHERE Toulouse and MUSE data Marseille V. Future developments 2

Observation with Adaptive Optics Atmosphere, Telescope Blurred Object AO system, CCD Noisy Image WFS CCD ● Residual turbulence with AO still ● Observation of satellites and asteroids produces “blurring“ ● Adaptive Optics (AO) partially corrects ● Photon and detector read-out noise the atmospheric turbulence Context ● Better images required (noiseless, ● AO greatly improves imaging Issues performances sharp edges, visible structures) Post-processing required to further improve image quality 3

Deconvolution Atmosphere, Telescope Blurred Object AO system, CCD Noisy Star Image WFS Observed PSF CCD Estimated object Deconvolution MISTRAL → Noise statistics → A priori on object Criterion to minimize - Observed PSF may significantly differ from the real system PSF - Errors on estimated object 4 Data fidelity A priori

Myopic deconvolution Atmosphere, Telescope Blurred Object AO system, CCD Noisy Star Image WFS Observed PSF CCD Estimated object and PSF Deconvolution MISTRAL → Noise statistics → A priori on object Criterion to minimize → PSD on the PSF - Better results than fixed-PSF - High number of unknowns 5 A priori Myopic term Data fidelity

Solution: parametrize the PSF VLT / SPHERE VLT / MUSE 500ηm 600ηm 700ηm 800ηm 900ηm Data Moffat MPHD MPHD AO correction Turbulence Diffraction & Strehl 6 quality Kolmogorov static artifacts PSF modelling

State of the art PSF from star t t+Δt Asteroid Star Protected PSF Data Estimated object Protected Deconvolution Image Data MISTRAL →Deconvolutional issues: Image VLT - SPHERE 7 mismatch true PSF (linked to object) / observed PSF (star) Credit: P. Vernazza (LAM)

State of the art Myopic deconvolution t t+Δt Asteroid Star Estimated object PSF Protected Data Protected Data Myopic Statistics on the PSF Deconvolution variations Estimated PSF MISTRAL Image →Deconvolutional issues 8 →Low statistical contrast: data (N pixels) << unknown (2 x N)

Goals of the thesis Parametric myopic deconvolution t Asteroid Estimated object Protected Image Data Protected Parametric Data Myopic Statistics on Deconvolution the parameters Estimated PSF variations MISTRAL Fried Strehl Parametric PSF Jitter... Telemetry →Suitable PSF for deconvolution → Physical meaning of the PSF’s parameters →High statistical contrast: data (N pixels) ≈ unknown (N+10) 9 → No need for on-sky PSF observation: astronomical & military usage

Work objectives ● Evaluate improvements of ● Error maps on estimated deconvolution using objects parametric PSF ● Stability toward parameters ● Myopic deconvolution Protected Data ● Apply method on data – Simulations – VLT (asteroids) – ONERA (satellites) 10

Toulouse 11