Detection of Subsurface Corrosion Beneath Coatings Using Thermal Time-Series Analysis (C2026-00421)

Qingyu ren, Hui Wang, Sreelakshmi Sreeharan, Alexander Chattos

Detecting localized corrosion concealed beneath polymer coatings on pipes and storage tanks is difficult with vision sensors alone. We introduce a non‑destructive testing (NDT) approach that builds a thermal 3D point cloud by tightly fusing data from monochrome light image derived geometry with transient thermography. A co‑calibrated monochrome–IR rig reconstructs a dense, metrically accurate point cloud, while a controlled halogen excitation induces a transient thermal response captured by the IR camera. The thermal sequence is rigidly registered to the 3D surface so that every point carries both geometric attributes (e.g., curvature, weld proximity, thickness proxies) and thermal descriptors (peak temperature rise, cooling rate, phase lag). We then perform point‑cloud segmentation to extract regions of interest (ROIs) and fuse geometric and thermal kinetics within each ROI to estimate a corrosion severity level using a physics‑informed, learning‑based classifier. The outcome is a spatially indexed map that localizes, visualizes, and quantifies under‑coating damage, enabling fast, non‑contact, and high‑resolution integrity assessment of large industrial assets such as above-ground pipelines and storage tanks.