Micro-computed tomography (micro-CT) is an imaging technique to examine the material's inner microstructure in detail (down to microscale) and even predict mechanical properties. The main limitation of micro-CT is a small field of view for high-resolution scans: only small and often non-representative specimens can be scanned to obtain detailed microstructure. In addition, high-resolution images take more time and are not resistant to x-ray artifacts. One solution is to synthetically increase the spatial resolution of scans in post-processing and remove defects by inpainting. Increasing resolution (super-resolution) techniques are widely used in different fields and have evolved rapidly by using deep learning algorithms. This work investigates deep learning-based methodologies and algorithms for super-resolution for micro-CT images of composite materials, its application for enabling automated fiber breaks detection for low resolution images.

Radmir Karamov completed his MSc in Skoltech, CMT, where he worked on the analysis of composite materials micro-CT. Now he continues his research as a joint PhD student in Skoltech and KU Leuven and works on applying machine learning algorithms in the field of composite materials.