Extending Effective Dynamic Range of Hyperspectral Line Cameras for Short Wave Infrared Imaging.

Area of Science:

• Optics and Photonics
• Computer Vision
• Materials Science

Background:

• Hyperspectral imaging offers rich spectral information but is limited by camera dynamic range (DR).
• Scenes with high DR can exceed the capture capabilities of standard short-wave infrared (SWIR) hyperspectral line cameras.
• This limitation hinders accurate material analysis in challenging applications.

Purpose of the Study:

• To propose and validate a multi-exposure method for increasing the DR of SWIR hyperspectral imaging.
• To assess the performance improvement in plastic detection within food waste and polymer sorting.
• To compare classification accuracy between the proposed high-dynamic-range (HDR) method and conventional single-exposure imaging.

Main Methods:

• Developed a multi-exposure technique to capture hyperspectral data across varying light intensities.
• Experimentally determined the DR of the SWIR camera and test samples.
• Applied Principal Component Analysis (PCA) for feature extraction and Support Vector Machine (SVM) classifiers for material identification.

Main Results:

• Increased the DR of SWIR hyperspectral imaging from 43 dB to 73 dB (SNR >= 20 dB).
• Achieved higher classification accuracies using the HDR method: 98% and 90% for food waste, and 95% for polymer sorting.
• Demonstrated superior performance of the multi-exposure HDR approach over single-exposure methods.

Conclusions:

• The proposed multi-exposure method effectively expands the DR of SWIR hyperspectral imaging.
• HDR hyperspectral imaging significantly enhances the accuracy of material classification tasks.
• This technique is valuable for applications requiring precise material identification under varying illumination conditions.