Small-amplitude vibrations are important in many engineering systems, including civil infrastructure, mechanical equipment, industrial assets, towers, stadia, transmission structures and tall buildings. These vibrations often contain useful information about structural condition, dynamic behaviour and early damage, but they are usually too small to be seen by the human eye or measured easily over large areas. Existing vibration-monitoring systems mainly rely on accelerometers, strain gauges and displacement sensors. These sensors provide accurate local measurements, but they are expensive to install, intrusive, spatially limited and difficult to deploy on large or hard-to-access assets.
This project will develop a new vision sensor for measuring imperceptible vibrations using ordinary video cameras. A standard camera will record the target structure or system, and computer-vision software will amplify small, normally invisible motions. These vibration patterns will then be converted into useful engineering information, including natural frequencies, mode shapes, damping estimates and damage indicators.
The project focuses on real-world sensing rather than motion magnification alone. It will address the practical barriers that currently limit field use, including camera shaking, changing lighting, outdoor noise, distance, scale calibration and uncertainty in the results. Civil infrastructure monitoring will be used as the primary validation domain, with testing from laboratory structures to at least one real civil structure and reference sensors used for validation. The goal is to produce a field-ready vision sensor that can support inspection, maintenance planning and rapid post-event assessment.
Aim
The project aims to develop and field-validate a new vision sensor capable of measuring imperceptible vibrations using ordinary video cameras. The sensor will use motion magnification to extract measurable vibration information from video and convert it into decision-grade engineering diagnostics. The project seeks to close the gap between laboratory demonstrations and real-world deployment. It will produce a framework that can take video of an operating structure or engineering system and return calibrated modal parameters, uncertainty estimates and actionable damage-detection results. Civil infrastructure will serve as the primary application and validation domain, with emphasis on supporting engineering decisions for asset owners, transport authorities and infrastructure managers.
Objectives
- Develop the core motion-magnification engine for the vision sensor. The project will develop and benchmark phase-based and learning-based motion magnification methods for small-amplitude vibration measurement. The module will be optimised for practical video conditions, including different camera specifications, distances and frame rates. It will aim to improve signal quality while reducing visual and measurement artefacts.
- Make the sensor robust for field deployment. The project will develop methods to compensate for camera motion, changing illumination and outdoor environmental noise. It will also include pixel-to-physical calibration so that video measurements can be converted into engineering units. This will allow video-based measurements to be compared with conventional contact sensors.
- Produce quantitative modal information with uncertainty. The project will convert magnified video and phase signals into natural frequencies, mode shapes and damping estimates. It will also quantify uncertainty in the measurements, rather than reporting single values only. The results will be benchmarked against accelerometers or other reference instrumentation.
- Develop application-focused vibration analytics. The project will define damage-sensitive features from full-field modal data, such as frequency shifts, mode-shape curvature changes and spatial residuals. These features will be used to detect, localise and estimate the severity of structural damage in the civil infrastructure validation domain. The methods will be validated on laboratory specimens with controlled and reversible damage scenarios.
- Demonstrate the vision sensor on a real structure. The project will deploy the end-to-end sensing framework on at least one real civil structure. The field study will assess accuracy, repeatability, environmental sensitivity, cost, practicality and ease of deployment. The sensor will be evaluated against reference instrumentation where available.
- Produce deployment guidelines and reusable tools. The project will produce practical guidelines for using motion-magnified video as a quantitative vibration sensor. These will cover camera placement, recording conditions, calibration, uncertainty reporting and interpretation of vibration-based indicators. The project will also release code, data and benchmark results where possible to support future research and industry adoption.
Significance
The project will involve collaboration with Curtin’s School of Civil and Mechanical Engineering to ensure strong alignment with real structural engineering needs and practical infrastructure-monitoring applications.
The significance of the project is both practical and scientific. Practically, it will establish a new sensing approach that transforms ordinary cameras into quantitative vibration-measurement devices. Unlike conventional contact sensors, the proposed vision sensor can provide non-contact and full-field measurements over large areas without requiring physical installation on the monitored asset. This makes the approach especially useful for rapid inspections, temporary monitoring and post-event assessment after storms, floods, vehicle impacts, vessel impacts or seismic events.
For asset owners and infrastructure managers, the project can support earlier damage detection, better maintenance planning and safer operation of critical infrastructure. It may reduce the need for extensive sensor installation and allow more frequent condition assessment of structures that are difficult, expensive or unsafe to instrument.
Scientifically, the project will advance the translation of motion magnification from qualitative visualisation to quantitative vibration sensing. It will deliver a calibrated video-based modal-identification pipeline, uncertainty-aware measurement methods, validated damage-detection tools and field-tested deployment guidance. The project will also build capability at the intersection of structural dynamics, computer vision and signal processing, positioning the candidate and host group in the growing area of intelligent, vision-based sensing for engineering systems.
Ideal Candidate
We are looking for a self-motivated HDR applicant with strong problem-solving, analytical and project management skills. The ideal candidate will have a background in civil/mechanical engineering, structural dynamics, computer vision, signal processing, applied mathematics, or related fields. Experience with Python or MATLAB, video/image processing, vibration analysis, machine learning, or structural health monitoring would be highly desirable. The candidate should be willing to conduct both computational research and practical laboratory/field testing. Strong communication skills and the ability to work across engineering and data-driven research areas are also important. Additionally, the applicants should meet the eligibility criteria for entry into a PhD program at Curtin University.
This project is open to International and Domestic applicants.
Scholarship
If you are identified as the preferred candidate for this project, you may be considered for an RTP scholarship.
Enquires and How to Apply
For enquires about this opportunity contact Dr Qian Li at Qli@curtin.edu.au
To formally apply submit an Expression of Interest to Dr Qian Li during the Central Scholarship round (July 1st – July 31st 2026)