Meshless Interpolation-Based Surface Thermography Characterization for Heat Source Detection

초록

Thermography via distributed sensors is essential in applications ranging from structural health monitoring to medical diagnostics, where localized thermal anomalies reveal subsurface damage or disease. Unlike dense sensor arrays, sparse sensor networks offer advantages such as gas exchange, mechanical conformability, and scalability. However, interpolated temperature maps are sensitive to sensor distribution and interpolation strategy. Many prior methods rely on application-specific data such as heat source location or variograms and lack a generalizable framework for comparing sensor density across scenarios. In this study, we pair a meshless radial basis function-based algorithm, the adaptive meshless approximation (AMA), with a normalized region-of-interest sensor density metric ( N-ROI ) to predict interpolation error without requiring prior knowledge of the region of interest (ROI). AMA yields temperature fields consistent with diffusive physics, enabling realistic reconstruction. We evaluate performance across gridded and random sampling, multiple phantom geometries, and 3-D surfaces. Results show that NROI strongly correlates with interpolation error across conditions, providing a scalable framework for designing low-density thermal sensor networks in diverse applications.

키워드