Abstract:Ground-based solar image restoration is a computationally expensive procedure that involves nonlinear optimization techniques. The presence of atmospheric turbulence produces perturbations in individual images that make it necessary to apply blind deconvolution techniques. These techniques rely on the observation of many short exposure frames that are used to simultaneously infer the instantaneous state of the atmosphere and the unperturbed object. We have recently explored the use of machine learning to accelerate this process, with promising results. We build upon this previous work to propose several interesting improvements that lead to better models. As well, we propose a new method to accelerate the restoration based on algorithm unrolling. In this method, the image restoration problem is solved with a gradient descent method that is unrolled and accelerated aided by a few small neural networks. The role of the neural networks is to correct the estimation of the solution at each iterative step. The model is trained to perform the optimization in a small fixed number of steps with a curated dataset. Our findings demonstrate that both methods significantly reduce the restoration time compared to the standard optimization procedure. Furthermore, we showcase that these models can be trained in an unsupervised manner using observed images from three different instruments. Remarkably, they also exhibit robust generalization capabilities when applied to new datasets. To foster further research and collaboration, we openly provide the trained models, along with the corresponding training and evaluation code, as well as the training dataset, to the scientific community.

Learning deconvolutions for astronomical images

Blind Deconvolution Using an Improved L0 Sparse Representation

Ground-to-air Aircraft Infrared Image Deblurring Based on Imaging Degradation Simulation

PI-AstroDeconv: A Physics-Informed Unsupervised Learning Method for Astronomical Image Deconvolution

A General Purpose Image Restoration Method with Deep Neural Network and Active Learning

Neural Network Based Point Spread Function Deconvolution For Astronomical Applications

Large Field Astronomical Image Restoration and Super-resolution Reconstruction Using Deep Learning

Image Restoration with Point Spread Function Regularization and Active Learning

Astronomical Image Restoration Using Variational Bayesian Blind Deconvolution

Astronomical Image Restoration and Point Spread Function Estimation with Deep Neural Networks

Blind Deconvolution with Principal Components Analysis for Wide-Field and Small-Aperture Telescopes

A New Deep Learning Method for Image Deblurring in Optical Microscopic Systems

Image restoration for synthetic aperture systems with a non-blind deconvolution algorithm via a deep convolutional neural network

Application of Physics-Informed Neural Networks in Removing Telescope Beam Effects

Data--driven Image Restoration with Option--driven Learning for Big and Small Astronomical Image Datasets

Point Spread Function Deconvolution Using a Convolutional Autoencoder for Astronomical Applications

Multi-Frame Blind Deconvolution Algorithm Based on Fourier-Constrained

Accelerating Multiframe Blind Deconvolution via Deep Learning

Restoration of Turbulence Degraded Images Using Improved Multi-Frame Blind Deconvolution Algorithm

A Universal Remote Sensing Image Quality Improvement Method with Deep Learning

Multi-scale Convolutional Neural Networks for Non-blind Image Deconvolution.