Super-Resolution Using a Deep Convolutional Network

Abstract

This master thesis deals with Super Resolution (SR) which is a set of image processing techniques used in computer vision to improve the quality of degraded images We focus on the differences between conventional image interpolation algorithms and deep learning based algorithms that have made significant progress in image quality improvement technology We will implement Conventional interpolation techniques and then we will implement deep learning algorithms SRCNN, VDSR,DWSR and EDSR Then the comparative study is performed in terms of calculating the peak signal-to-noise ratio PSNR and the structural similarity index SSIM.