Dabaja, Hassan (Aix-Marseille University), Noura, Hassan (Aix-Marseille University), Ouladsine, Mustapha (LIS Laboratory (UMR CNRS 7020), Aix-Marseille University, 13397 )

Vision-Based Structural Health Monitoring: A Survey

Scheduled for presentation during the Regular Session "Fault diagnosis II" (WeBC), Wednesday, June 11, 2025, 14:40−15:00, Room B

33rd Mediterranean Conference on Control and Automation, June 10-13, 2025, Tangier, Morocco

This information is tentative and subject to change. Compiled on May 9, 2025

Abstract

The field of Vision-based Structural Health Monitoring (SHM) utilizes computer vision (CV), imaging techniques, and machine learning to quantify the integrity of structures. This survey sheds light on this field presenting an overview over key techniques and advancements. In particular the survey presents classical image processing, digital image correlation (DIC), 3D modeling, and deep learning methods. Edge detection, thresholding, and segmentation can be considered as the classical damage identification tools, whereas DIC provides accurate deformation and strain measurements. Recent advancements in 3D modeling and deep learning have further boosted the efficiency and accuracy of crack detection, structural evaluation, and monitoring. Advanced deep learning methods such as convolutional neural networks (CNNs) and transfer learning (TL) have proven their effectiveness in automating crack detection and segmentation tasks under challenging and noisy conditions. This survey outlines strengths, weaknesses, and applicability to SHM with a view to transforming SHM into a non-contact, inexpensive, and fully automated monitoring system.