Near-Infrared Spectroscopy Non-Destructive Detection Modeling for Starch Content in Kernels of 58 Rainfed Corn...

Area of Science:

• Agricultural Science
• Spectroscopy
• Chemometrics

Background:

• Traditional starch content determination in corn is inefficient and destructive.
• Developing rapid, non-destructive methods is crucial for the corn industry.

Purpose of the Study:

• To develop a rapid, non-destructive method for determining corn starch content using near-infrared hyperspectral imaging.
• To identify key wavelengths and build predictive models for accurate starch quantification.

Main Methods:

• Near-infrared hyperspectral imaging applied to 58 corn varieties.
• Spectral preprocessing included wavelet transform, multiplicative scatter correction, and standard normal variate transformation.
• Wavelength selection used competitive adaptive reweighted sampling and sparrow search algorithm optimization.
• Four predictive models (PLS, ANN, CNN, GBDT) were compared.

Main Results:

• Identified 14 key wavelengths (1020.65-1647.71 nm) strongly correlated with starch content.
• Artificial Neural Network (ANN) model showed the best performance with R²=0.826, RMSE=0.759%, and RPD=2.40.
• The selected wavelengths have clear chemical bond assignments and physical interpretability.

Conclusions:

• Near-infrared hyperspectral imaging offers a viable non-destructive method for corn starch determination.
• The developed method supports corn quality grading, breeding, and raw material screening.
• Key wavelengths identified can facilitate the development of portable detection instruments.