An Overview of Image Super-resolution Reconstruction
Yanqiong Shen, Tao Jiang, Chao Zhang
Abstract
In recent years, image super-resolution (SR) technology has been one of the research hotspots in the field of computer vision and image processing, aiming at converting low-resolution images or videos into high-resolution to obtain clearer and more realistic visual effects. In this paper, we comprehensively review the current key algorithms in the field of image super-resolution. In terms of algorithms, we not only review the classical interpolation methods, such as bilinear interpolation, bicubic interpolation, and nearest-neighbor interpolation algorithms but also delve into deep learning-based image super-resolution methods. The rapid development of deep learning has driven the rapid progress of image super-resolution techniques, which covers the application of cutting-edge technologies such as convolutional neural networks (CNN) and generative adversarial networks (GAN).