Classification of oat and groat kernels using NIR hyperspectral imaging.

Area of Science:

• Agricultural Science
• Analytical Chemistry
• Spectroscopy

Background:

• Oat groats (hull-less kernels) are a key quality indicator in the oat market.
• Accurate classification of oat types is crucial for quality control and processing.

Purpose of the Study:

• To develop and validate an innovative procedure for classifying oat and groat kernels.
• To assess the feasibility of using hyperspectral imaging (HSI) and chemometrics for this classification task.

Main Methods:

• Acquired hyperspectral images of oat and groat samples in the near-infrared (NIR) range (1006-1650 nm).
• Utilized Principal Component Analysis (PCA) for exploratory data analysis.
• Applied Partial Least Squares-Discriminant Analysis (PLS-DA) to build classification models.
• Selected optimal wavelengths (1132, 1195, 1608 nm) using a bootstrap-VIP procedure for faster processing.

Main Results:

• Achieved near 100% prediction accuracy in classifying single oat and groat kernels using HSI.
• Demonstrated that classification models can be effectively built using only three specific wavelengths.
• The developed method is objective and non-destructive.

Conclusions:

• Hyperspectral imaging coupled with chemometrics provides a highly accurate method for oat and groat kernel classification.
• The identification of key wavelengths enables rapid and efficient classification for industrial applications.
• This HSI-based approach offers significant potential for quality control and innovative sorting solutions in the oat industry.