“…(18) corresponds to the Wiener filtering [Ref. 1, chap. 4] with a non-null prior mean bold-italicmbold-italicoo^(f)=h˜*(f)i˜(f)+SnSo(f)m˜o(f)|h˜(f)|2+SnSo(f).…”
Section: Results On Simulated Astronomical Datamentioning
confidence: 99%
“…However, the correction is partial and residual blurring remains, impacting high spatial frequencies of the observed object. Therefore, the observation system includes post-processing to restore the high frequencies 1 …”
.Adaptive optics (AO) corrected image restoration is particularly difficult, as it suffers from the lack of knowledge on the point spread function (PSF) in addition to usual difficulties. An efficient approach is to marginalize the object out of the problem and to estimate the PSF and (object and noise) hyperparameters only, before deconvolving the image using these estimates. Recent works have applied this marginal myopic deconvolution method, based on the maximum a posteriori estimator, combined with a parametric model of the PSF, to a series of AO-corrected astronomical and satellite images. However, this method does not enable one to infer global uncertainties on the parameters. We propose a PSF estimation method, which consists in choosing the minimum mean square error estimator and computing the latter as well as the associated uncertainties thanks to a Markov chain Monte Carlo algorithm. We validate our method by means of realistic simulations, in both astronomical and satellite observation contexts. Finally, we present results on experimental images for both applications: an astronomical observation on Very Large Telescope/spectro-polarimetric high-contrast exoplanet research with the Zimpol instrument and a ground-based LEO satellite observation at Côte d’Azur Observatory’s 1.52 m telescope with Office National d'Etudes et de Recherches Aérospatiales’s ODISSEE AO bench.
“…(18) corresponds to the Wiener filtering [Ref. 1, chap. 4] with a non-null prior mean bold-italicmbold-italicoo^(f)=h˜*(f)i˜(f)+SnSo(f)m˜o(f)|h˜(f)|2+SnSo(f).…”
Section: Results On Simulated Astronomical Datamentioning
confidence: 99%
“…However, the correction is partial and residual blurring remains, impacting high spatial frequencies of the observed object. Therefore, the observation system includes post-processing to restore the high frequencies 1 …”
.Adaptive optics (AO) corrected image restoration is particularly difficult, as it suffers from the lack of knowledge on the point spread function (PSF) in addition to usual difficulties. An efficient approach is to marginalize the object out of the problem and to estimate the PSF and (object and noise) hyperparameters only, before deconvolving the image using these estimates. Recent works have applied this marginal myopic deconvolution method, based on the maximum a posteriori estimator, combined with a parametric model of the PSF, to a series of AO-corrected astronomical and satellite images. However, this method does not enable one to infer global uncertainties on the parameters. We propose a PSF estimation method, which consists in choosing the minimum mean square error estimator and computing the latter as well as the associated uncertainties thanks to a Markov chain Monte Carlo algorithm. We validate our method by means of realistic simulations, in both astronomical and satellite observation contexts. Finally, we present results on experimental images for both applications: an astronomical observation on Very Large Telescope/spectro-polarimetric high-contrast exoplanet research with the Zimpol instrument and a ground-based LEO satellite observation at Côte d’Azur Observatory’s 1.52 m telescope with Office National d'Etudes et de Recherches Aérospatiales’s ODISSEE AO bench.
The use of IEEE 754-2008 half-precision floatingpoint numbers is an emerging trend in Graphical Processing Units' architecture. Being such a compact way of representing data, its use may speed up programs by reducing the memory bandwidth usage and allowing hardware designers to fit more computing units within the same die space. In this paper, we highlight the acceleration offered by the use of half floatingpoint numbers over different implementations of the same operation, a 2D convolution. We show that even though it may lead up to a significant speed-up, the degradation brought by this new format is not always negligible. Then, we choose a deconvolution problem inspired by the SKA radio-telescope processing pipeline to show how half floats behave in a more complex application.
scite is a Brooklyn-based organization that helps researchers better discover and understand research articles through Smart Citations–citations that display the context of the citation and describe whether the article provides supporting or contrasting evidence. scite is used by students and researchers from around the world and is funded in part by the National Science Foundation and the National Institute on Drug Abuse of the National Institutes of Health.