Resolution enhancement of cementitious microstructure images and phases quantification using deep learning
Resolution enhancement of cementitious microstructure images and phases quantification using deep learning
This paper introduces a deep learning method for reconstructing and segmenting cement paste microstructural images in Backscattered Electron (BSE) mode. Using a dataset of 2400 full-scale BSE images (258,818 after cropping), the Local Implicit Image Function (LIIF) super-resolution network enhanced image resolution up to 30 ×, reducing noise and improving microstructural detail. This improved the accuracy of segmentation, which was carried out using the SegFormer network. SegFormer outperformed traditional methods like U-Net and DeepLabv3 + in terms of mean Intersection over Union (MIoU) and detail preservation. The segmentation results quantified porosity, unhydrated cement, and hydration products, aligning closely with Mercury Intrusion Porosimetry (MIP) and Quantitative X-Ray Diffraction (QXRD) measurements. The hydration degree from BSE segmentation also matched well with Thermogravimetric Analysis (TGA) calculations. The method showed strong generalization capabilities, effectively handling diverse BSE images. This study confirms the method's reliability and accuracy for concrete microstructure analysis.
Backscattered electron image, Cement microsctructure and composition, Deep learning, Phase quantification, Segmentation, Super-resolution
Ma, Yiming
ec7c6106-8faa-40be-b3d1-5b1a1c330cd7
Qian, Hanjie
7ebd2a2e-335f-4b30-8b38-76add7a93791
Zou, Dujian
f932d3d9-b218-4268-a86e-0bb63aec1e31
Zhou, Ao
5b42c2a4-26b2-416e-ab3c-446f1ece7a20
Liu, Tiejun
07e72a65-be75-4b13-b54d-9ed949c93470
Li, Ye
86d13351-982d-46c3-9347-22794f647f86
11 January 2025
Ma, Yiming
ec7c6106-8faa-40be-b3d1-5b1a1c330cd7
Qian, Hanjie
7ebd2a2e-335f-4b30-8b38-76add7a93791
Zou, Dujian
f932d3d9-b218-4268-a86e-0bb63aec1e31
Zhou, Ao
5b42c2a4-26b2-416e-ab3c-446f1ece7a20
Liu, Tiejun
07e72a65-be75-4b13-b54d-9ed949c93470
Li, Ye
86d13351-982d-46c3-9347-22794f647f86
Ma, Yiming, Qian, Hanjie, Zou, Dujian, Zhou, Ao, Liu, Tiejun and Li, Ye
(2025)
Resolution enhancement of cementitious microstructure images and phases quantification using deep learning.
Construction and Building Materials, 461, [139909].
(doi:10.1016/j.conbuildmat.2025.139909).
Abstract
This paper introduces a deep learning method for reconstructing and segmenting cement paste microstructural images in Backscattered Electron (BSE) mode. Using a dataset of 2400 full-scale BSE images (258,818 after cropping), the Local Implicit Image Function (LIIF) super-resolution network enhanced image resolution up to 30 ×, reducing noise and improving microstructural detail. This improved the accuracy of segmentation, which was carried out using the SegFormer network. SegFormer outperformed traditional methods like U-Net and DeepLabv3 + in terms of mean Intersection over Union (MIoU) and detail preservation. The segmentation results quantified porosity, unhydrated cement, and hydration products, aligning closely with Mercury Intrusion Porosimetry (MIP) and Quantitative X-Ray Diffraction (QXRD) measurements. The hydration degree from BSE segmentation also matched well with Thermogravimetric Analysis (TGA) calculations. The method showed strong generalization capabilities, effectively handling diverse BSE images. This study confirms the method's reliability and accuracy for concrete microstructure analysis.
Text
Revised manuscript-submit clean
- Accepted Manuscript
Restricted to Registered users only
Request a copy
More information
Accepted/In Press date: 6 January 2025
e-pub ahead of print date: 11 January 2025
Published date: 11 January 2025
Keywords:
Backscattered electron image, Cement microsctructure and composition, Deep learning, Phase quantification, Segmentation, Super-resolution
Identifiers
Local EPrints ID: 507761
URI: http://eprints.soton.ac.uk/id/eprint/507761
ISSN: 0950-0618
PURE UUID: 2d917474-573a-417c-a76f-75f17a06c9b2
Catalogue record
Date deposited: 06 Jan 2026 11:07
Last modified: 08 Jan 2026 03:27
Export record
Altmetrics
Contributors
Author:
Yiming Ma
Author:
Hanjie Qian
Author:
Dujian Zou
Author:
Ao Zhou
Author:
Tiejun Liu
Author:
Ye Li
Download statistics
Downloads from ePrints over the past year. Other digital versions may also be available to download e.g. from the publisher's website.
View more statistics