Submit

Login Login

Register Register

Home / Journals / SDHM / Online First / doi:10.32604/sdhm.2026.073282
Journal Menu
Special Issues
Table of Content

All issues

Online First

2026

2025

2024

2023

2022

2021

2020

2019

2018

2017

2014

2013

2012

2011

2010

2009

2008

2007

2006

2005

Open Access

ARTICLE

Concrete Bridge Defect Monitoring and Quantitative Identification via U-Net and Mathematical Morphology

Caiping Huang*, Yulong Mei, Wangyuan Tian, Zihang Yu
School of Civil Engineering, Architecture and Environment, Hubei University of Technology, Wuhan, China
* Corresponding Author: Caiping Huang. Email: email
(This article belongs to the Special Issue: Smart Sensors and Smart CFRP Components for Structural Health Monitoring of Aerospace, Energy and Transportation Structures)

Structural Durability & Health Monitoring https://doi.org/10.32604/sdhm.2026.073282

Received 15 September 2025; Accepted 26 November 2025; Published online 06 July 2026

Abstract

Bridge damage detection is critical to bridge maintenance practices. However, traditional inspection methods are plagued by high labour intensity and low operational efficiency. To enhance the intelligence, objectivity, and efficiency of bridge damage detection, this study proposes an automated approach for the identification and quantitative measurement of concrete defects. Specifically, this method adopts the Visual Geometry Group (VGG) network as the backbone of the U-Net architecture to perform semantic segmentation on images containing typical concrete defects, including spalling, cracks, and exposed reinforcement bars. Subsequently, mathematical morphology algorithms are employed to optimise the segmented images, thereby mitigating errors induced by semantic segmentation. Utilising MATLAB software, the area (or length) of concrete defects is quantified by referencing objects with known dimensions. To validate the proposed method, semantic segmentation experiments were conducted on defect images collected from 17 in-service reinforced concrete bridges in Xuchang City, Henan Province. Experimental results indicate that the VGG16-U-Net model not only accurately locates and classifies three common concrete defects under complex background conditions but also enables the quantification of their area or length. The obtained results fully meet the requirements of practical engineering inspections. For the identification of concrete spalling, cracks, and exposed reinforcement, the model achieved a Mean Pixel Accuracy (MPA) of 90.53% and a Mean Intersection over Union (MIoU) of 80.54%. Furthermore, the application of mathematical morphology to optimize segmentation errors further improved the accuracy of defect quantitative calculation, with the optimized absolute error ranging from 0.08% to 0.21%.

Keywords

Concrete defects; deep learning; U-Net; mathematical morphology; quantitative calculation

  • 131

    View

  • 19

    Download

  • 0

    Like

Related articles
Copyright© 2026 Tech Science Press
© 1997-2026 TSP (Henderson, USA) unless otherwise stated

Share Link