Large-view-field astronomical image restoration and super-resolution reconstruction method based on deep learning

The invention belongs to the field of image processing, and particularly relates to a large-view-field astronomical image restoration and super-resolution reconstruction method based on deep learning, and the method comprises the steps: S1, making a large-view-field astronomical image data set; s2, constructing a Resmodel by constraining the restoration task and the super-resolution task at the same time, wherein the model comprises a feature extraction module, a feature correction module and a profile generation module; and S3, performing end-to-end image restoration and super-resolution reconstruction by using the Resmodel: extracting features of different resolutions by a feature extraction module, then correcting error parts of the image features by a feature correction module, and fusing the corrected features with the corresponding resolution features to generate a restored image and a super-resolution image. According to the method, the high-quality restored image and the super-resolution image can be