Survey of single image super-resolution reconstruction

• Published in: IET image processing Vol. 14; no. 11; pp. 2273 - 2290
• Main Authors: Li, Kai, Yang, Shenghao, Dong, Runting, Wang, Xiaoying, Huang, Jianqiang
• Format: Journal Article
• Language: English
• Published: The Institution of Engineering and Technology 18.09.2020; Wiley
• Subjects: computer vision; deep neural network; deep‐level network; feature extraction; GAN‐based deep learning method; generative adversarial network; hierarchical feature‐based model; hyperspectral imaging; image input information; image processing; image reconstruction; image resolution; interpolation; learning (artificial intelligence); LR images; LR–HR images; medical imaging; multiscale super‐resolution reconstruction; network model; neural nets; or multiple images; original image processing‐based model; popular CNN‐based deep learning method; remote sensing; Review Article; single image super‐resolution reconstruction; super‐resolution images; generative adversarial network; LR images; multiscale super-resolution reconstruction; network model; feature extraction; learning (artificial intelligence); image resolution; deep neural network; deep-level network; hierarchical feature-based model; image processing; original image processing-based model; remote sensing; image reconstruction; popular CNN-based deep learning method; interpolation; single image super-resolution reconstruction; LR–HR images; computer vision; super-resolution images; GAN-based deep learning method; or multiple images; hyperspectral imaging; neural nets; medical imaging; image input information
• ISSN: 1751-9659; 1751-9667
• DOI: 10.1049/iet-ipr.2019.1438

Image super-resolution reconstruction refers to a technique of recovering a high-resolution (HR) image (or multiple images) from a low-resolution (LR) degraded image (or multiple images). Due to the breakthrough progress in deep learning in other computer vision tasks, people try to introduce deep neural network and solve the problem of image super-resolution reconstruction by constructing a deep-level network for end-to-end training. The currently used deep learning models can divide the SISR model into four types: interpolation-based preprocessing-based model, original image processing based model, hierarchical feature-based model, and high-frequency detail-based model, or shared the network model. The current challenges for super-resolution reconstruction are mainly reflected in the actual application process, such as encountering an unknown scaling factor, losing paired LR–HR images, and so on.