Blueberry bruise non-destructive detection based on hyperspectral information fusion combined with multi-strategy...

Area of Science:

• Agricultural Science
• Computer Vision
• Spectroscopy

Background:

• Mechanical damage significantly impacts fruit market value.
• Early detection of damage is vital for agricultural management.
• Non-destructive methods are preferred for assessing fruit quality.

Purpose of the Study:

• To develop a non-destructive method for early detection of mechanical damage in blueberries.
• To integrate hyperspectral image fusion with advanced machine learning models.
• To optimize a Support Vector Machine (SVM) model for enhanced accuracy.

Main Methods:

• Hyperspectral image fusion was employed.
• Spectral features were extracted using the Successive Projections Algorithm (SPA).
• Image features were extracted using the Grey Level Co-occurrence Matrix (GLCM).
• Support Vector Machine (SVM) hyperparameters were optimized using a multi-strategy improved Beluga Whale Optimization (BWO) algorithm.
• Comparative analysis with Random Forest (RF) and Partial Least Squares Discriminant Analysis (PLS-DA) was performed.

Main Results:

• The optimized SVM model achieved the highest classification accuracy.
• The optimized SVM model demonstrated 95.00% accuracy on the test set.
• Feature fusion of hyperspectral data proved effective for detecting blueberry bruising.

Conclusions:

• The optimized SVM model shows significant potential for post-harvest blueberry assessment.
• Hyperspectral imaging combined with SVM offers an efficient approach for early bruising detection.
• Further research is needed to validate the model across diverse environmental conditions and fruit types.