Nondestructive Detection for Egg Freshness Based on Hyperspectral Scattering Image Combined with Ensemble Learning

Area of Science:

• Agricultural Science
• Spectroscopy
• Food Science

Background:

• Nondestructive testing (NDT) methods are crucial for quality assessment in the food industry.
• Hyperspectral imaging offers a promising avenue for objective and rapid quality evaluation.
• Egg freshness is a key quality attribute influenced by biochemical changes over time.

Purpose of the Study:

• To develop and validate a hyperspectral imaging method for accurate egg freshness detection.
• To investigate the influence of light source incident angles on detection accuracy.
• To elucidate the underlying mechanisms of scattering hyperspectral technology in freshness assessment.

Main Methods:

• Utilized scattering hyperspectral technology for egg freshness analysis.
• Applied spectral preprocessing and feature wavelength extraction.
• Employed stacking ensemble learning with weak classifiers to enhance accuracy.
• Evaluated different hyperspectral modes: scattering, reflection, transmission, and mixed.

Main Results:

• Achieved a maximum accuracy of 100% for egg freshness detection using scattering hyperspectral imaging.
• Scattering hyperspectral mode outperformed reflection, transmission, and mixed modes.
• Detection accuracy was inversely proportional to the incident angle of the light source.
• Smaller incident angles allowed for greater collection of scattering light, rich in biochemical information.

Conclusions:

• Scattering hyperspectral technology, particularly with optimized incident angles, offers a highly accurate NDT method for egg freshness.
• The findings provide critical insights for improving NDT accuracy and selecting optimal light source parameters.
• This technology holds significant potential for real-time, online non-destructive quality assessment of eggs.