[Detection of Hawthorn Fruit Defects Using Hyperspectral Imaging]

Area of Science:

• Agricultural Science
• Food Science
• Spectroscopy

Background:

• Hawthorn fruit quality is crucial for consumption and processing.
• Traditional defect detection methods can be destructive and time-consuming.
• Distinguishing between different defects like bruises, insect damage, calyx, and stem-end is challenging with standard imaging.

Purpose of the Study:

• To develop and validate a hyperspectral imaging technique for non-destructive detection of defects in hawthorn fruit.
• To identify key spectral features and wavelengths indicative of specific hawthorn fruit defects.
• To establish a robust model for accurate classification of damaged and intact hawthorn fruits.

Main Methods:

• Hyperspectral imaging (380-1000 nm) was used to collect data from 230 hawthorn fruits with various defects.
• Spectral data preprocessing included Standardized Normal Variate (SNV) transformation.
• Partial Least Squares-Discriminant Analysis (PLS-DA) and Least Squares-Support Vector Machine (LS-SVM) models were employed for classification.
• Principal Component Analysis (PCA) and image processing algorithms ('Sobel', 'Regiongrow') were used for feature extraction.

Main Results:

• The SNV pretreatment combined with PLS-DA identified ten important wavelengths for defect detection.
• The LS-SVM model achieved a classification accuracy of 91.23% for defect detection.
• Image analysis using PCA and specific algorithms demonstrated high detection precision for bruises (95.65%) and insect damage (86.67%).

Conclusions:

• Hyperspectral imaging is a viable non-destructive technology for detecting defects in hawthorn fruit.
• The identified key wavelengths and established models provide a theoretical basis for real-time defect detection systems.
• This approach enhances quality control in hawthorn fruit production and processing.