Galaxy Image Restoration with Shape Constraint

Nammour, Fadi; Schmitz, Morgan A.; Mboula, Fred Maurice Ngolè; Starck, Jean-Luc; Girard, Julien N.

doi:10.1007/s00041-021-09880-9

Cited by 3 publications

(15 citation statements)

References 32 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…While training the U-Net for the DL methods with and without shape constraint, we normalized the pixel values of the input images by 4×10 3 for the optical case and by 2×10 3 for the radio case in order to make their magnitudes close to unity, so that the activation functions in the neural network could better discriminate data. We compare the deep learning methods to Sparse Reconstruction Algorithm (SRA) and Shape COnstraint REstoration algorithm (SCORE) (Nammour et al 2021), and additionally CLEAN for the radio case. SRA is a deconvolution method based on sparsity and positivity, while SCORE is its extension that uses an additional shape constraint, as described in Nammour et al (2021).…”

Section: Numerical Experiments and Resultsmentioning

confidence: 99%

“…These filters allow to capture the anisotropy of the galaxy image and are used in the constraint as a set of windows such that at least one them reduces the noise in the image and emphasize the useful signal. We have also shown that adding such a constraint to a sparse deconvolution approach reduces both the shape and pixel errors (Nammour et al 2021). An alternative could have been to directly use the ellipticity in the loss function rather than our shape constraint.…”

Section: The Shape Constraintmentioning

confidence: 93%

“…where ω i j i, j are non-negative scalar weights (their computation is detailed in Nammour et al (2021)) and ψ j j are K directional and multi-scale filters, derived from a curvelet-like decomposition (Starck et al 2015;Kutyniok & Labate 2012). These filters allow to capture the anisotropy of the galaxy image and are used in the constraint as a set of windows such that at least one them reduces the noise in the image and emphasize the useful signal.…”

Section: The Shape Constraintmentioning

confidence: 99%

“…where x is the Tikhonov filter output of the observed image, M is the shape constraint and γ ∈ R + its weight parameter as introduced in Nammour et al (2021). The choice of the hyperparameters is detailed in B.1 and B.2.…”

Section: Shapenet Deconvolutionmentioning

confidence: 99%

“…These great results can be explained by the fact that DL learns, or rather approximates, the mean of the posterior distribution of the solution. As there is no guarantee that a non-linear deconvolution process preserves the galaxies' shapes, Nammour et al (2021) introduced a new shape penalization term and showed that adding this penalty to sparse regularization improves both the solution shape and the MSE. In this paper, we propose a new deconvolution method, called ShapeNet, by extending the Tikhonet method to include the shape constraints.…”

Section: Introductionmentioning

confidence: 99%

See 4 more Smart Citations

ShapeNet: Shape Constraint for Galaxy Image Deconvolution

Nammour,

Akhaury,

Girard

et al. 2022

Preprint

View full text Add to dashboard Cite

Deep Learning (DL) has shown remarkable results in solving inverse problems in various domains. In particular, the Tikhonet approach is very powerful to deconvolve optical astronomical images (Sureau et al. 2020). Yet, this approach only uses the 2 loss, which does not guarantee the preservation of physical information (e.g. flux and shape) of the object reconstructed in the image. In Nammour et al. ( 2021), a new loss function was proposed in the framework of sparse deconvolution, which better preserves the shape of galaxies and reduces the pixel error. In this paper, we extend Tikhonet to take into account this shape constraint, and apply our new DL method, called ShapeNet, to optical and radio-interferometry simulated data set. The originality of the paper relies on i) the shape constraint we use in the neural network framework, ii) the application of deep learning to radio-interferometry image deconvolution for the first time, and iii) the generation of a simulated radio data set that we make available for the community. A range of examples illustrates the results.

show abstract

Section: Numerical Experiments and Resultsmentioning

confidence: 99%

Section: The Shape Constraintmentioning

confidence: 93%

Section: The Shape Constraintmentioning

confidence: 99%

Section: Shapenet Deconvolutionmentioning

confidence: 99%

Section: Introductionmentioning

confidence: 99%

See 3 more Smart Citations

ShapeNet: Shape Constraint for Galaxy Image Deconvolution

Nammour,

Akhaury,

Girard

et al. 2022

Preprint

View full text Add to dashboard Cite

show abstract

ShapeNet: Shape constraint for galaxy image deconvolution

Nammour

Akhaury

Girard

et al. 2022

A&A

View full text Add to dashboard Cite

Deep learning (DL) has shown remarkable results in solving inverse problems in various domains. In particular, the Tikhonet approach is very powerful in deconvolving optical astronomical images. However, this approach only uses the 2 loss, which does not guarantee the preservation of physical information (e.g., flux and shape) of the object that is reconstructed in the image. A new loss function has been proposed in the framework of sparse deconvolution that better preserves the shape of galaxies and reduces the pixel error. In this paper, we extend the Tikhonet approach to take this shape constraint into account and apply our new DL method, called ShapeNet, to a simulated optical and radio-interferometry dataset. The originality of the paper relies on i) the shape constraint we use in the neural network framework, ii) the application of DL to radio-interferometry image deconvolution for the first time, and iii) the generation of a simulated radio dataset that we make available for the community. A range of examples illustrates the results.

show abstract

The SAMI Galaxy Survey: the difference between ionized gas and stellar velocity dispersions

Colless

D’Eugenio

et al. 2022

Monthly Notices of the Royal Astronomical Society

View full text Add to dashboard Cite

We investigate the mean locally-measured velocity dispersions of ionised gas (σgas) and stars (σ*) for 1090 galaxies with stellar masses log (M*/M⊙) ≥ 9.5 from the SAMI Galaxy Survey. For star-forming galaxies, σ* tends to be larger than σgas, suggesting that stars are in general dynamically hotter than the ionised gas (asymmetric drift). The difference between σgas and σ* (Δσ) correlates with various galaxy properties. We establish that the strongest correlation of Δσ is with beam smearing, which inflates σgas more than σ*, introducing a dependence of Δσ on both the effective radius relative to the point spread function and velocity gradients. The second-strongest correlation is with the contribution of active galactic nuclei (AGN) (or evolved stars) to the ionised gas emission, implying the gas velocity dispersion is strongly affected by the power source. In contrast, using the velocity dispersion measured from integrated spectra (σap) results in less correlation between the aperture-based Δσ (Δσap) and the power source. This suggests that the AGN (or old stars) dynamically heat the gas without causing significant deviations from dynamical equilibrium. Although the variation of Δσap is much smaller than that of Δσ, a correlation between Δσap and gas velocity gradient is still detected, implying there is a small bias in dynamical masses derived from stellar and ionised gas velocity dispersions.

show abstract

Galaxy Image Restoration with Shape Constraint

Cited by 3 publications

References 32 publications

ShapeNet: Shape Constraint for Galaxy Image Deconvolution

ShapeNet: Shape Constraint for Galaxy Image Deconvolution

ShapeNet: Shape constraint for galaxy image deconvolution

The SAMI Galaxy Survey: the difference between ionized gas and stellar velocity dispersions

Contact Info

Product

Resources

About