Computational Imaging Method for Thermal Infrared Hyperspectral Imaging Based on a Snapshot Divided-Aperture System

Area of Science:

• Optics and Photonics
• Computational Imaging
• Spectroscopy

Background:

• High spectral resolution and snapshot imaging are challenging to achieve simultaneously in thermal infrared spectral imaging.
• Existing methods often compromise on speed or spectral detail.

Purpose of the Study:

• To develop a computational imaging method for high spectral resolution thermal infrared imaging with snapshot capability.
• To enable hyperspectral reconstruction from low-dimensional multispectral measurements for dynamic scenarios.

Main Methods:

• A self-developed divided-aperture snapshot multispectral camera captures nine low-spectral-resolution images in a single exposure.
• Star-point array calibration ensures precise registration of sub-channel images.
• A neural network model reconstructs 127-channel hyperspectral information using a dataset acquired by a Fourier-transform infrared hyperspectral camera (FTIR HCam).

Main Results:

• The proposed method successfully reconstructs hyperspectral information from multispectral measurements.
• The system maintains compactness and snapshot imaging capability.
• Experimental results validate the effectiveness for dynamic thermal infrared sensing.

Conclusions:

• The developed method offers a viable technical approach for hyperspectral sensing in dynamic thermal infrared scenarios.
• It overcomes the limitations of simultaneous snapshot imaging and high spectral resolution.
• This advancement supports applications requiring rapid, detailed spectral analysis in the thermal infrared range.