Determination of hardness for maize kernels based on hyperspectral imaging

Area of Science:

• Agricultural Science
• Spectroscopy
• Data Science

Background:

• Maize kernel hardness is a critical quality attribute.
• Accurate and non-destructive hardness measurement is challenging.
• Hyperspectral imaging offers potential for rapid quality assessment.

Purpose of the Study:

• To develop a quantitative method for non-destructive hardness detection in maize kernels.
• To establish a predictive model for maize kernel hardness using hyperspectral imaging.
• To validate the accuracy and reliability of the proposed method.

Main Methods:

• Establishing a regression model for hardness and moisture content.
• Utilizing partial least squares regression (PLSR) with characteristic wavelengths selected by successive projection algorithm (SPA).
• Applying reflectance hyperspectral imaging in the 399.75-1005.80 nm range.

Main Results:

• A predictive model for moisture content was developed.
• The final hardness prediction model achieved a coefficient of determination (R²) of 0.912.
• Validation metrics included RMSE of 17.76 MPa, RPD of 3.41, and RER of 14.

Conclusions:

• Hyperspectral imaging provides an effective method for rapid, non-destructive hardness detection in maize kernels.
• The developed model demonstrates high accuracy and reliability for quality control.
• This approach can significantly improve post-harvest processing and quality assessment of maize.