Single image super-resolution via deep learning
Lists
Only you see this
Abstract
Single image Super-resolution (SISR) is a computer vision (CV) problem that aims to acquire a high-resolution (HR) image from a distorted low-resolution (LR) image, making it a valuable technology that could be utilized in various fields such as photography, medical imaging, satellite imaging, etc. As a result of the advancement of computing hardware and richer computational power, deep learning-based image super-resolution models have emerged at an unprecedented rate. This paper reviews SISR and its recent development. Three widely used deep architectures: convolutional neural network (CNN), generative adversarial network (GAN), and transformer are explained. Next, six different deep learning-based models that summarize research on SISR are analyzed. Finally, this review concludes with applications of SR, current challenges SISR models encountered, and potential future research directions.
BibTeX
@inproceedings{cao2022single,
author = {Zhenyue Cao and Xuan Liu and Zhenkun Wang},
title = {{Single image super-resolution via deep learning}},
booktitle = {IEEE/CVF International Conference on Computer Vision},
year = {2022},
doi = {10.1109/cvidliccea56201.2022.9824576},
}Loading related content...