Mixed-Scale Dense Convolutional Neural Network based Improvement of Glass Fiber-reinforced Composite CT Images — ICTMS Committee Dep of Applied Maths Australian National University
  1. Schedule
  2. Advanced-materials e.g. composites, foams & engineered tissue
  3. Theme 3: Advanced-materials e.g. composites & Timber
  4. Mixed-Scale Dense Convolutional Neural Network based Improvement of Glass Fiber-reinforced Composite CT Images

Mixed-Scale Dense Convolutional Neural Network based Improvement of Glass Fiber-reinforced Composite CT Images (132)

Tim Elberfeld 1 , Shabab Bazrafkan 1 , Jan De Beenhouwer 1 , Jan Sijbers 1
  1. University of Antwerp, Antwerpen, FLANDERS, Belgium

For the study of glass fiber-reinforced polymers (GFRP),μCT is the method of choice. ObtainingGFRP parameters from aμCT scan is difficult, due to the presence of noise and artifacts. We propose a methodto improve GFRP image quality using a recently introduced deep neural network. We describe the network’ssetup and the data generation and show how the trained network improves the reconstruction.

Full Paper