Deep Learning Based Image Enhancement using Super Resolution

Abstract

With the ever-growing field of microscopy there is pretty much a necessity of high - resolution microscopic images. A microscope may have powerful magnifying lenses, but if the resolution is poor, the magnified image is just blur and no useful insights can be gained from such images. Traditional techniques like Structured Illumination Microscopy (SIM) are not feasible enough for proper use and current solutions based on deep learning assume that the input image is noise free. Based on our research and existing applications related to deep learning-based image enhancement, our proposed solution of deep learning based General Adversarial Network (GAN), will help jointly denoise and super-resolved microscopy images. Thus, this project has competitive applications in different research areas including biomedical microscopy, medical diagnosis, astronomical research, surveillance or investigation, etc., and many other areas as well.