Super-resolution image reconstruction and applications

Abstract

As technology continues to advance, many image processing methods are being developed and improved upon. As capturing photos becomes easier and more researchers look to take advantage of computer vision in their fields, Super-Resolution becomes a significant topic for study. It will prove to be useful not only for public consumers but also for any researchers who utilise digital imaging in their experiments. This project aims to provide an analysis of the existing options for Super-Resolution while pinpointing a practical approach to utilize and improve on them. In the process, Super-Resolution using Convolutional Neural Network was found to be one of the more practical ways to improve the quality of an image. Thus, an approach to make use of Convolutional Neural Network on top of traditional Super-Resolution methods of Bicubic Interpolation was tested and proven to be effective.