Acquisition and Analysis of Hyperspectral Thermal Images for Sample Segregation

Area of Science:

• Optics and Photonics
• Materials Science
• Spectroscopy

Background:

• Conventional thermography struggles with accurate temperature determination due to unknown material emissivities.
• Fourier transform infrared spectroscopy-based imagers are complex and may not be suitable for all applications.

Purpose of the Study:

• To introduce and evaluate a novel, simpler hyperspectral thermal imager operating in the 8.0–14.0 µm range.
• To demonstrate the capability of hyperspectral thermal imaging to differentiate materials based on their unique spectral signatures.

Main Methods:

• Development of a hyperspectral thermal camera using a low-order scanning Fabry-Pérot interferometer.
• Acquisition of 3D hyperspectral data cubes for various materials including Vantablack, painted aluminum, borosilicate glass, Kapton tape, and bare aluminum.
• Application of Principal Component Analysis (PCA) and logistic regression for data analysis and material discrimination.

Main Results:

• Hyperspectral thermal imaging clearly segregated different material samples, with borosilicate glass being the most distinguishable.
• PCA effectively reduced data dimensionality, retaining 92% of variance with significantly fewer spectral bands.
• Logistic regression achieved high true positive rates (up to 99% for borosilicate glass) in classifying materials, significantly outperforming standard thermography.

Conclusions:

• The developed hyperspectral thermal camera offers a simpler and more informative alternative to conventional thermography.
• Material-specific spectral signatures captured by the hyperspectral imager enable accurate sample discrimination.
• This technology holds potential for applications requiring precise material identification in the thermal spectrum.