Deep-learning image enhancement and fibre segmentation from time-resolved computed tomography of fibre-reinforced composites
Guo, Rui ; Stubbe, Johannes ; Zhang, Yuhe LU ; Schlepütz, Christian Matthias ; Gomez, Camilo Rojas ; Mehdikhani, Mahoor ; Breite, Christian ; Swolfs, Yentl and Villanueva-Perez, Pablo LU
(2023)
In Composites Science and Technology
244.
- Abstract
Monitoring the microstructure and damage development of fibre-reinforced composites during loading is crucial to understanding their mechanical properties. Time-resolved X-ray computed tomography enables such an in-situ, non-destructive study. However, the photon flux and fibre-matrix contrast limit its achievable spatial and temporal resolution. In this paper, we push the limits of temporal and spatial resolution for the microstructural analysis of unidirectional continuous carbon fibre-reinforced epoxy composites by establishing a new pipeline based on CycleGAN for unsupervised super-resolution and denoising and U-Net-id for individual fibre segmentation. After illustrating the benefits of a 3D CycleGAN over a 2D one, we show that... (More)
Monitoring the microstructure and damage development of fibre-reinforced composites during loading is crucial to understanding their mechanical properties. Time-resolved X-ray computed tomography enables such an in-situ, non-destructive study. However, the photon flux and fibre-matrix contrast limit its achievable spatial and temporal resolution. In this paper, we push the limits of temporal and spatial resolution for the microstructural analysis of unidirectional continuous carbon fibre-reinforced epoxy composites by establishing a new pipeline based on CycleGAN for unsupervised super-resolution and denoising and U-Net-id for individual fibre segmentation. After illustrating the benefits of a 3D CycleGAN over a 2D one, we show that data enhanced by this pipeline can yield similar segmentation quality to that of a slow-acquisition, high-quality scan that took up to 200 times longer to acquire. This pipeline, therefore, enables more robust data extraction from fast time-resolved X-ray tomography, removing a critical stumbling block for this technique.
(Less)
- author
- Guo, Rui
; Stubbe, Johannes
; Zhang, Yuhe
LU
; Schlepütz, Christian Matthias
; Gomez, Camilo Rojas
; Mehdikhani, Mahoor
; Breite, Christian
; Swolfs, Yentl
and Villanueva-Perez, Pablo
LU
- organization
- publishing date
- 2023-11-10
- type
- Contribution to journal
- publication status
- published
- subject
- keywords
- A. Carbon fibre, A. Polymer-matrix composites, D. Non-destructive testing, D. X-ray computed tomography, Deep learning
- in
- Composites Science and Technology
- volume
- 244
- article number
- 110278
- publisher
- Elsevier
- external identifiers
-
- scopus:85159364498
- ISSN
- 0266-3538
- DOI
- 10.1016/j.compscitech.2023.110278
- language
- English
- LU publication?
- yes
- additional info
- Funding Information: We acknowledge the Paul Scherrer Institut, Villigen, Switzerland for the provision of synchrotron radiation beamtime at the TOMCAT beamline X02DA of the SLS. R. Guo would like to thank F.H Wagner for the discussion of U-Net-id and acknowledge his PhD scholarship from China Scholarship Council (202006430010). C. Breite and M. Mehdikhani would like to acknowledge FWO Flanders for their postdoctoral fellowships, COCOMI (1231322N) and ToughImage (1263421N), respectively. Funding Information: We acknowledge the Paul Scherrer Institut, Villigen, Switzerland for the provision of synchrotron radiation beamtime at the TOMCAT beamline X02DA of the SLS. R. Guo would like to thank F.H Wagner for the discussion of U-Net-id and acknowledge his PhD scholarship from China Scholarship Council ( 202006430010 ). C. Breite and M. Mehdikhani would like to acknowledge FWO Flanders for their postdoctoral fellowships, COCOMI ( 1231322N ) and ToughImage (1263421N), respectively. Publisher Copyright: © 2023 Elsevier Ltd
- id
- c7d1ad93-6fd6-404d-b5cc-5490925e6608
- date added to LUP
- 2024-01-15 09:08:52
- date last changed
- 2024-01-15 09:08:52
@article{c7d1ad93-6fd6-404d-b5cc-5490925e6608,
abstract = {{<p>Monitoring the microstructure and damage development of fibre-reinforced composites during loading is crucial to understanding their mechanical properties. Time-resolved X-ray computed tomography enables such an in-situ, non-destructive study. However, the photon flux and fibre-matrix contrast limit its achievable spatial and temporal resolution. In this paper, we push the limits of temporal and spatial resolution for the microstructural analysis of unidirectional continuous carbon fibre-reinforced epoxy composites by establishing a new pipeline based on CycleGAN for unsupervised super-resolution and denoising and U-Net-id for individual fibre segmentation. After illustrating the benefits of a 3D CycleGAN over a 2D one, we show that data enhanced by this pipeline can yield similar segmentation quality to that of a slow-acquisition, high-quality scan that took up to 200 times longer to acquire. This pipeline, therefore, enables more robust data extraction from fast time-resolved X-ray tomography, removing a critical stumbling block for this technique.</p>}},
author = {{Guo, Rui and Stubbe, Johannes and Zhang, Yuhe and Schlepütz, Christian Matthias and Gomez, Camilo Rojas and Mehdikhani, Mahoor and Breite, Christian and Swolfs, Yentl and Villanueva-Perez, Pablo}},
issn = {{0266-3538}},
keywords = {{A. Carbon fibre; A. Polymer-matrix composites; D. Non-destructive testing; D. X-ray computed tomography; Deep learning}},
language = {{eng}},
month = {{11}},
publisher = {{Elsevier}},
series = {{Composites Science and Technology}},
title = {{Deep-learning image enhancement and fibre segmentation from time-resolved computed tomography of fibre-reinforced composites}},
url = {{http://dx.doi.org/10.1016/j.compscitech.2023.110278}},
doi = {{10.1016/j.compscitech.2023.110278}},
volume = {{244}},
year = {{2023}},
}