Learning to denoise astronomical images with U-nets

Abstract

Astronomical images are essential for exploring and understanding the universe. Optical telescopes capable of deep observations, such as the Hubble Space Telescope, are heavily oversubscribed in the Astronomical Community. Images also often contain additive noise, which makes de-noising a mandatory step in post-processing the data before further data analysis. In order to maximise the efficiency and information gain in the post-processing of astronomical imaging, we turn to machine learning. We propose Astro U-net, a convolutional neural network for image de-noising and enhancement. For a proof-of-concept, we use Hubble space telescope images from WFC3 instrument UVIS with F555W and F606W filters. Our network is able to produce images with noise characteristics as if they are obtained with twice the exposure time, and with minimum bias or information loss. From these images, we are able to recover $95.9\%$ of stars with an average flux error of $2.26\%$. Furthermore the images have, on average, 1.63 times higher signal-to-noise ratio than the input noisy images, equivalent to the stacking of at least 3 input images, which means a significant reduction in the telescope time needed for future astronomical imaging campaigns.

Super-resolving Herschel imaging: a proof of concept using Deep Neural Networks

“…Supervised learning requires the use of a training dataset with known truth values. Two separate training sets were used to train the network presented in this paper: (i) a simulated training set made using images generated by a modified version of the Empirical Galaxy Generator (EGG) software (Schreiber et al 2017) as both the the target and input images, and (ii) using the JCMT SCUBA-2 450 µm maps from the STUDIES project (Wang et al 2017) as target examples, combined with the Herschel SPIRE maps for the COSMOS field as input images (Levenson et al 2010;Viero et al 2013). Table 1 shows the FWHM, confusion limit and pre-interpolation pixel scales of the instruments used in this paper.…”

“…Similar networks have been used to enhance and remove noise from astronomical images at other wavelengths (e.g. Vojtekova et al 2020).…”

“…Characteristics of the Herschel SPIRE and JCMT SCUBA-2 instruments. Confusion limits and PSF widths are fromNguyen et al (2010), Wang et al (2017 andDempsey et al (2013).…”

MaLeFiSenta: Machine Learning for FilamentS Identification and Orientation in the ISM

“…Recent developments in the field of deep learning, where object segmentation networks such as U-Net [28], Mask R-CNN (Region-based Convolution Neural Network) [29], FastFCN (Fully Convolution Network) [30], Gated-SCNN (Gated-Shape Convolution Neural Network) [31], DeepLab [32] provide an efficient and fast image segmentation results. Moreover, neural networks have already showed their efficiency in improving astronomical data and solve the problem of noise [33,34].…”

Encoding large-scale cosmological structure with generative adversarial networks

“…Developing a meaningful latent encoding space for data can have several applications, such as semisupervised classification, disentangling style and content of images, unsupervised clustering, and dimensionality reduction (Hinton & Salakhutdinov 2006), as can be seen, for example, in the case of variational autoencoders (Makhzani et al 2015) and adversarial autoencoders (Kingma & Welling 2013). In the case of astrophysics or cosmology, AEs could be used to help remove instrumental or astrophysical signal contamination (e.g., point sources, beam, andinstrumental noise) (Vojtekova et al 2020) or for inpainting masked areas while preserving the statistics of the data (Sadr & Farsian 2020;Puglisi & Bai 2020).…”