Review on Video Super Resolution: Methods and Metrics

Abstract

Video Super Resolution (VSR) is a wide area in image processing, where numerous research works are going on. Super Resolution (SR) means the upstaging of Low Resolution (LR) frames to High Resolution (HR) frames with minimal loss in image quality. The goal of this paper is to give the comprehensive review of various VSR methods like Convolutional Neural Network (CNN), Generative Adversarial Network (GAN), Gated Recurrent Unit (GRU), Spatial Temporal Transformer (STT), Fractional-Sine Cosine Algorithm (F-SCA), Recurrent Back-Projection Network (RBPN), Temporally-Deformable Alignment Network (TDAN). Finally, the readers will get more ideas about VSR based methods and metrics.