Session: RIP: Predictive Modeling and Characterization of Corrosion Processes in Complex Environments (In Honor of Professor Digby Macdonald) (Part IV of IV)
Thermal 3D Point Clouds for Corrosion Monitoring (RIP2026-00047)
Characterizing corrosion processes in complex environments requires more than traditional 2D imaging, which often fails to capture the geometric depth of an object. This research proposes a non-invasive evaluation framework utilizing thermal 3D point clouds derived from Structure from Motion (SfM) techniques. By integrating thermal data with 3D geometric reconstruction, we demonstrate a method to detect and monitor environmental precursors to corrosion, such as thermal bridges indicating insulation failure or localized heat concentration, on non-planar surfaces. Our technique to make thermal 3D point clouds makes use of a dual-camera system of a high quality RGB camera and a thermal camera. An RGB point cloud is created using existing SfM techniques and software. Then, using camera parameters that are collected through co-calibration of both cameras, we find a projective matrix that can project the temperature values onto the RGB pixels, and by extension the RGB point cloud. Using a camera array system, SfM data can be captured instantaneously and can be taken over a long period of time, allowing for a 3D point cloud time series infused with thermal data. Ultimately, this modeling technique helps monitor corrosion with 3D spatial awareness, allowing engineers to more accurately gauge the quality of a surface.
