Astronomical images allow people to explore the Universe and monitor space; however, due to the long distances involved, such images are generally collected using telescopic equipment. The equipment optical characteristics and the imaging environment cause image degradation, such as blurring, lost details, and sometimes serious losses of object structures and contours, thus limiting the applications of these images. Unfortunately, improving the equipment to acquire much sharper images is expensive. Therefore, we propose a post-processing structure learning method to restore astronomical images that is low in cost but has exciting effects. The proposed method uses single backbone neural networks or their simple combinations to solve a series of image restoration problems, including point spread function (PSF) estimation, non-blind deconvolution, and blind deconvolution. In tests on simulated and real astronomical images, the proposed method achieves dramatic improvements compared to other state-of-the-art methods. Although this work concentrates on astronomical images, the proposed framework is applicable to a wide range of fields.

Saliency, the distinctive parts of an image, has shown good potential for many applications (e.g., image interpretation and target detection). In this letter, a novel saliency detection method based on background context-aware is proposed for synthetic aperture radar (SAR) images. According to a statistical analysis of SAR image characteristics, several reference background patches (RBPs) are selected. Then, the dissimilarities between the current patch and the RBPs are used to calculate the local saliency, which is further enhanced for the final saliency under the multiscale framework. Experimental results demonstrate the effectiveness of the proposed method, which outperforms some state-of-the-art methods.

This paper focuses on point-spread function (PSF) estimation for astronomical images containing only single extended objects captured by adaptive optics systems. The problem is very different from, and much more challenging than, PSF estimation with point-like source images. We propose a new PSF estimation framework based on deep-learning technology. In this framework, PSFs can be estimated “end-to-end” using the original degraded images. Moreover, such a framework can precisely address different sources of blur without requiring accurate prior information about the PSF, image or imaging system. Therefore, the method is practical. We test the proposed method on both simulated and real data, and the favorable results show that the method is valid and performs much better than classical methods do.

The existing astronomical image restoration and superresolution reconstruction methods have problems such as low efficiency and poor results when dealing with images possessing large fields of view. Furthermore, these methods typically only handle fixed-size images and require step-by-step processing, which is inconvenient. In this paper, a neural network called Res&RecNet is proposed for the restoration and superresolution reconstruction of astronomical images with large fields of view for direct imaging instruments. This network performs feature extraction, feature correction, and progressive generation to achieve image restoration and superresolution reconstruction. The network is constructed using fully convolutional layers, allowing it to handle images of any size. The network can be trained using small samples and can perform image restoration and superresolution reconstruction in an end-to-end manner, resulting in high efficiency. Experimental results show that the network is highly effective in terms of processing astronomical images with complex scenes, generating image restoration results that improve the peak signal-to-noise ratio (PSNR) and structural similarity index measure (SSIM) by 4.69 (dB)/0.073 and superresolution reconstruction results that improve the PSNR and SSIM by 1.97 (dB)/0.077 over those of the best existing algorithms, respectively.

This study introduces PSDetNet, an innovative deep neural network tailored for the autonomous detection of pointlike astronomical sources by leveraging feature-matching techniques. PSDetNet comprises two primary modules: feature extraction and matching localization. The feature extraction module is built on residual blocks and adopts an encoder–decoder framework to distill features from images robustly. The matching localization module employs a patch-by-patch comparison against a preconstructed template, which is crafted through the alignment and weighted aggregation of numerous exemplar pointlike source samples, capturing the quintessential distribution characteristics of pointlike sources. The experimental results demonstrate that this network can accurately detect pointlike sources in astronomical imagery with high purity and completeness. It operates end to end and uses a fully convolutional architecture that allows for flexible processing of images of any size. This ability considerably enhances its applicability across various practical scenarios.