Super-resolution through Machine Learning

The aims of this study are two-fold: Aim 1: Establish a protocol and imaging procedure to collect series of images in multiple resolutions, modalities, noise, and atmospheric conditions to evaluate the performance and robustness of super-resolution algorithms. Aim 2: Identify, implement, and compare two algorithms representative of two super-resolution method families: model-based methods and learning-based methods. The outcome of the project will be an evaluation database, a quantitative comparison pipeline, and implementation of reference algorithms. Together, they will form the basis for the development and evaluation of future super-resolution methods.
Idiap Research Institute
Jul 01, 2020
Jan 31, 2021