Remote sensing images and medical imaging are essential for many applications, but their quality can usually be degraded due to limitations in imaging technology and complex imaging environments. To address this, various imaging methods have been developed to restore sharp and high-quality images from degraded observational data. However, most traditional model-based imaging methods usually require predefined hand-crafted prior assumptions, which are difficult to handle in complex applications. On the other hand, deep learning-based methods are often considered as black boxes, lacking transparency and interpretability. In this talk, we first present a new blind deblurring learning framework and high-order progressive SR network that utilizes alternating iterations of shrinkage thresholds. Then, we present a novel dual-domain unified framework that offers a great deal of flexibility for multi-sparse-view CT reconstruction through a single model. Experimental results on real and synthetic remote sensing image and medical image datasets demonstrate the superiority of our methods compared to existing deblurring methods.