Meat species authentication using portable hyperspectral imaging.

Area of Science:

• Food Science
• Analytical Chemistry
• Spectroscopy

Background:

• Meat species fraud poses significant risks to consumers and food safety.
• Hyperspectral imaging (HSI) offers a powerful method for detecting meat adulteration by combining spectral and spatial data.

Purpose of the Study:

• To develop a portable hyperspectral imager (HSI) for on-site meat adulteration detection.
• To create a robust discrimination model with improved generalization capabilities across different instruments.

Main Methods:

• Designed a portable push broom HSI with 5 nm spectral and 0.1 mm spatial resolution, controlled by Raspberry Pi.
• Implemented a model transfer method to ensure model sharing and correct for instrument variations.
• Utilized a Support Vector Machine (SVM) classifier with Spectral Space Transformation (SST).

Main Results:

• The model transfer method successfully corrected spectral differences and enhanced model robustness.
• Achieved a highest accuracy of 94.91% for meat adulteration detection.
• Developed a visualization map to illustrate the distribution of meat adulteration.

Conclusions:

• The portable HSI facilitates rapid, on-site analysis for meat species authentication.
• The developed model transfer method provides a reliable solution for cross-instrument meat assessment.
• This technology is valuable for food safety and quality control applications.