Blind restoration of solar images via the Channel Sharing Spatio-temporal Network

Abstract

Context. Due to the presence of atmospheric turbulence, the quality of solar images tends to be significantly degraded when observed by ground-based telescopes. The adaptive optics (AO) system can achieve partial correction but stops short of reaching the diffraction limit. In order to further improve the imaging quality, post-processing for AO closed-loop images is still necessary. Methods based on deep learning (DL) have been proposed for AO image reconstruction, but the most of them are based on the assumption that the point spread function (PSF) is spatially invariant. Aims. Our goal is to construct clear solar images by using a sophisticated spatially variant end-to-end blind restoration network. Methods. The proposed channel sharing spatio-temporal network (CSSTN) consists of three sub-networks: a feature extraction network (FEN), channel sharing spatio-temporal filter adaptive network (CSSTFAN), and a reconstruction network (RN). First, CSST-FAN generates two filters adaptively according to features generated from three inputs. Then these filters are delivered to the proposed channel sharing filter adaptive convolutional (CSFAC) layer in CSSTFAN to convolve with the previous or current step features. Finally, the convolved features are concatenated as input of RN to restore a clear image. Ultimately, CSSTN and the other three supervised DL methods are trained on the binding real 705nm photospheric and 656nm chromospheric AO correction images as well as the corresponding speckle reconstructed images. Results. The results of CSSTN, the three DL methods, and one classic blind deconvolution method evaluated on four test sets are shown. The imaging condition of the first photospheric and second chromospheric set is the same as training set, except for the different time given in the same hour. The imaging condition of the third chromospheric and fourth photospheric set is the same as the first and second, except for the Sun region and time. Our method restores clearer images and performs best in both the peak signal-to-noise ratio (PSNR) and contrast among these methods.

Multi-frame blind deconvolution and phase diversity with statistical inclusion of uncorrected high-order modes

“…It can also be used for making high-quality quick-look movies. Wang et al (2021) similarly trained their NN with Specklerestored images. After training it could output images of similar quality much faster and with the additional feature that it does not assume spatially invariant blurring, so it has the potential to handle anisoplanatism better.…”

Improving the Spatial Resolution of Solar Images Based on an Improved Conditional Denoising Diffusion Probability Model

“…The methods based on the imaging model perform reconstruction tasks by obtaining the pointspread function (PSF), including speckle imaging, multiframe blind deconvolution (MFBD), phase diversity (PD), etc. The speckle imaging uses the statistical characteristics of atmospheric turbulence to reconstruct the phase and amplitude of the object (Wang et al 2021). This method has been continuously developed since it was proposed in 1970, and it has also been used on solar telescopes such as GST and New Vacuum Solar Telescope (NVST).…”

“…More information can be used to restore image details through MOMFBD with PD (Löfdahl et al 2021). This kind of method can obtain good image quality by using the information from multiple images and different prior assumptions or prior values for PSF or the observed image (Ramos et al 2018;Jia et al 2019), and has been widely used in SR reconstruction of solar images (Ramos & Olspert 2021;Wang et al 2021).…”

“…It trained the network with the collected system calibration data, and then estimated the PSF in any field of view from several discretely sampled star maps. Wang et al (2021) proposed a channelsharing spatiotemporal network, which used a complex spatial variant end-to-end blind restoration network to construct clear solar images. It is proved that this method cannot only remove the spatially variant blur included in the training, but also remove the unseen spatially variant blur, and has good generalization performance.…”

Multi-frame blind deconvolution and phase diversity with statistical inclusion of uncorrected high-order modes

“…Carlos Diaz Baso recently designed a version of this NN 1 , which is now installed at the SST, providing an almost real-time view of the target area during observations with a resolution that is similar to the science data delivered by the SSTRED pipeline (Löfdahl et al 2021); it can also be used for making high-quality quick-look movies. Wang et al (2021) similarly trained their NN with Specklerestored images. After training, it could output images of similar quality much faster and with the additional feature that it does not assume spatially invariant blurring, and so has the potential to handle anisoplanatism better.…”