Surface temperature determination using long range thermal emission spectroscopy based on a first order scanning...

Area of Science:

• Materials Science
• Optical Engineering
• Spectroscopy

Background:

• Surface temperature determination via thermography is challenging due to the coupled dependence on temperature and emissivity.
• The temperature-emissivity system is underdetermined without prior knowledge of material emissivity.

Purpose of the Study:

• To demonstrate material recognition using hyperspectral thermal imaging (HSTI).
• To address the underdetermined temperature-emissivity problem in thermography.

Main Methods:

• Acquisition of HSTI in the 8-14 µm range using a microbolometer and Fabry-Pérot interferometer.
• Application of a logistic regression model for material differentiation.
• Utilizing partial least squares (PLS) models for surface temperature prediction.

Main Results:

• Successful differentiation of polyimide tape, sapphire, borosilicate glass, fused silica, and alumina ceramic.
• Material recognition achieved with true positive rates exceeding 94% at temperatures as low as 34.0 ± 0.05 °C.
• Surface temperature prediction with a root mean square error (RMSE) of 1.10 °C, outperforming conventional thermography.

Conclusions:

• Hyperspectral thermal imaging enables accurate material identification based on spectral signatures.
• This approach provides a practical solution to the underdetermined temperature-emissivity problem.
• The developed method offers improved surface temperature prediction accuracy compared to conventional thermography.