Computer Vision for Performance Assessment of Infrastructure

Summary of the project

With the increasing trend of urbanisation and resulting further limitations on surface space, the use of underground space becomes a popular choice to deal with urban problems. Most major infrastructure schemes in the UK, for example, Crossrail, High Speed Two and Thames Tideway involve substantial tunnelling and excavations works in cities. However, utilising underground space, especially in the soft ground comes with an inevitable cost of causing ground movements during and after construction, and their effects on overlying and adjacent buildings become main concerns and issues of modern tunnelling (Mair, 1998). For this reason, rigorous and comprehensive monitoring and survey schemes on nearby buildings are required during urban underground construction.

The goal of this research is to provide accurate non-contact monitoring solutions through computer and machine vision techniques for masonry assets subjected to the ground movement using accessible tools in a less controlled environment. Specifically, the developed solutions do not require highly specialist hardware or stringent monitoring condition. The monitoring data can be captured through common practice using commercial laser scanners and consumer-grade digital cameras or come from archived laser scans and photographs of the asset of interest. Such solutions provide rich information in 3D about the asset that unavailable through traditional monitoring techniques and at the same time, accurate measurement results, which can be used directly in established assessment and decision-marking routine.

Aims and objectives

1. Developing a robust technique for accurate displacement monitoring in 3D of large masonry assets without stringent data acquisition requirements or intrusive instrumentations;
2. Providing a solution that makes automated measurements of cracking in photographs taken as part of standard condition surveys.

Novelty of the research methodology

1. Utilising state-of-the-art computer vision techniques for full-field displacement measurements of civil structures, which provides rich information in 3D about the asset that unavailable through traditional monitoring techniques and at the same time, accurate measurement results;
2. Developing a novel algorithm that addresses difficulties associated with the displacement analysis using point cloud;
3. Providing an innovative image-based crack monitoring solution, which is non-intrusive, more cost-effective and safer to use.

Any companies or collaborators involved

This DPhil studentship is co-funded by the UK Engineering and Physical Sciences Research Council (EPSRC) Industrial Cooperative Awards in Science & Technology (CASE) Award Scheme and Ove Arup & Partners Limited.
Ove Arup & Partners Limited will allow the student to attend one or more of its research establishments for a minimum of three months during and for the purpose of the project.

This project falls within the EPSRC Built Environment, Ground Engineering and Sensors & Instrumentation research areas within the Engineering Theme.