Access Official Technical PDF

Share Summary:LinkedInEmail

Vision-based displacement monitoring can help engineers measure movement in large structures where installing a conventional contact sensor is impractical. Suspension and cable-stayed bridge decks are a clear example: at midspan, movement can be substantial, while a nearby fixed point for a displacement transducer may not be available. A camera system observes the structure remotely, providing a non-contact approach to tracking vertical and transverse deck motion.

The central challenge is not simply to detect a target in an image. A useful measurement system must relate image coordinates to the physical structure, maintain reliable tracking as the deck moves, and account for both translation and rotation. Long viewing distances and the use of telescopic optics make calibration especially important.

Why measure bridge-deck displacement?

Structural health monitoring uses measurements to help identify changes in how infrastructure behaves during construction and throughout its service life. Displacement is one of the quantities of interest for a suspension or cable-stayed bridge. Observing how the deck moves can support assessment of its dynamic response and help reveal behaviour that warrants further investigation.

Contact-based displacement transducers typically depend on a stable reference near the point being monitored. That condition can be difficult to satisfy for a large bridge span. A remote vision system avoids direct attachment of a measuring device to the moving deck, but introduces its own measurement problems: cameras must be calibrated for the scene, and image-based observations must be converted into meaningful estimates of motion.

Vision-based displacement monitoring at long range

In a camera, a three-dimensional point is projected onto a two-dimensional image. The relationship depends on the camera’s optical and geometric parameters, as well as its position and orientation relative to the observed structure. Calibration establishes the information needed to interpret image measurements in physical space.

At long range, a camera fitted with a long-focal-length lens can behave differently from the idealized perspective model often used in close-range applications. The work described in the source develops a calibration approach suited to this setting, using an affine camera model. Its practical motivation is to avoid relying on an extensive set of known three-dimensional reference points. Instead, the method uses limited geometric information, including distances between targets, and is designed to allow in-situ calibration while the deck is moving.

This is relevant to field deployment because calibration needs to reflect the actual camera arrangement and observation conditions. For system integrators, calibration is not a preliminary administrative step: it establishes the foundation for turning pixel locations into defensible displacement estimates.

Tracking translation and rotation, not just image movement

A bridge deck can translate and rotate as it moves. If a measurement system treats every change in target position as simple translation, rotational motion may be misinterpreted as displacement. The developed tracking methodology therefore estimates the deck’s translational movement as well as its rotation.

The approach uses multiple cameras and targets to observe the moving structure from more than one viewpoint. The stated configuration requires a minimum of two cameras and four targets to estimate the motion. Tracking a set of targets helps describe the movement of the object as a whole, rather than relying on one image point that may not represent the deck’s complete motion.

A recursive estimator based on the Kalman filter framework supports the tracking process. In general terms, a recursive filter combines the current observations with an estimate of the system’s evolving state. Here, the motion model and image observations are used to follow the target arrangement over time. This is intended to support dynamic observation, where measurements need to be updated repeatedly as the structure moves.

Measurement objectives and validation

The system was developed for observing the evolution of vertical and transverse displacement in suspension and cable-stayed bridge decks. The stated objectives include a sampling rate above 20 Hz, movement with an amplitude of several metres, and accuracy better than 10 mm. These are design targets described for the measurement methodology; they should not be read as guaranteed performance for every camera installation or field condition.

The work includes numerical simulations and physical model tests, including outdoor and indoor experiments. These tests address both calibration and motion tracking. They provide a way to examine how the methods perform with different camera arrangements, target configurations and image-measurement noise. The source also discusses practical elements of an experimental setup, such as target support, camera synchronization and a reference measurement system.

For engineers considering a similar deployment, these details point to several system-level requirements:

  • Camera geometry: Position and orient cameras so the target arrangement can be observed from useful viewpoints.
  • Target visibility: Use targets that can be identified consistently in the acquired images.
  • Synchronization: Coordinate camera acquisition when observations from multiple views are combined.
  • Reference and calibration: Establish the geometry needed to interpret image coordinates and assess the resulting measurements.
  • Motion modelling: Account for rotation as well as translation when estimating deck movement.

Practical implications for infrastructure monitoring

Remote camera measurement is not a universal replacement for contact sensors. Its value is in applications where direct attachment or a nearby stable reference is difficult, and where the structure can be observed with suitable camera and target geometry. The system’s usefulness depends on the complete measurement chain: optics, calibration, image processing, tracking and synchronization all contribute to the final motion estimate.

For plant managers and technical decision-makers responsible for large assets, the broader lesson is that measurement design should start with the physical constraints of the structure. For system integrators, the bridge-deck example shows why long-range calibration and motion estimation need to be designed together rather than treated as separate tasks.

Conclusion

Vision-based displacement monitoring offers a non-contact route to observing movement in large structures where conventional displacement transducers may lack a suitable reference. The described approach combines long-range camera calibration with multi-camera target tracking and recursive estimation to capture deck translation and rotation. Its practical implications are clear: reliable results depend on calibration suited to the optics and scene, adequate target and camera geometry, and a measurement design that reflects the structure’s full motion.

Access Official Technical PDF

About The Author