Paper Contents
Abstract
This survey delves into the realm of image super-resolution (SR), a critical aspect for high-quality displays and medical imaging. Focusing on deep learning algorithms for single-image super-resolution (SISR), the survey categorizes and reviews recent SR techniques, analyzes performance-affecting factors, evaluates algorithms on the Urban100 dataset, and discusses challenges and future directions in the field. Built upon an examination of 12 existing review papers, the survey sheds light on the complexities of this ill-posed inverse problem. The study introduces innovative SR approaches, including SR3 using denoising diffusion, MSRN with multi-scale residual blocks, SAN with second-order channel attention, LapSRN employing a Laplacian Pyramid Network, and CinCGAN for unsupervised learning.
Copyright
Copyright © 2024 Sinchana P. This is an open access article distributed under the Creative Commons Attribution License.
