A Review of Super-Resolution Reconstruction Based on Supervised Learning

Abstract

With the infiltration of deep learning ideas into the field of super-resolution reconstruction, the image reconstruction effect based on deep neural network algorithm has been greatly improved compared with the traditional algorithm, and it has become the current research hotspot and mainstream direction. In this paper, the super-resolution reconstruction algorithm based on supervised learning is combed. Firstly, the background of image super-resolution reconstruction and three traditional super-resolution reconstruction algorithms are introduced from the degradation process. Then super-resolution reconstruction algorithms based on supervised learning are divided into four categories, which are linear shallow network, residual network, recursive network and densely connected network. Then the experimental part introduces the common data set of super-resolution reconstruction and two evaluation indexes PSNR and SSIM, and compares the performance of different algorithms. Finally, the research direction of super-resolution reconstruction algorithm based on supervised learning is prospected, and some problems worthy of further research are put forward.