Non-Destructive Detection of Abnormal Chicken Eggs by Using an Optimized Spectral Analysis System

Area of Science:

• Agricultural Science
• Spectroscopy
• Data Analysis

Background:

• Environmental and genetic factors cause internal chicken egg abnormalities, such as bloody or damaged yolks.
• Spectrometry offers potential for real-time detection, but measurement system optimization is underexplored.

Purpose of the Study:

• To optimize parameters for detecting internal egg abnormalities using visible and near-infrared (Vis/NIR) spectrometry.
• To investigate the impact of light sources, configuration, and sensor positions on detection performance.

Main Methods:

• Utilized Vis/NIR spectrometry (192-1110 nm) and multivariate data analysis.
• Developed a partial least-squares discriminant analysis (PLS-DA) model to classify normal and abnormal eggs.
• Applied band selection methods (WRC, SFS, SPA) to optimize spectral bands.

Main Results:

• Achieved a highest classification accuracy of 98.7% across various system parameters.
• Reduced spectral bands from 1028 to fewer than 7 using band selection techniques.
• Identified light source type and sensor/illumination configuration as critical factors.

Conclusions:

• The study successfully established optimal parameters for Vis/NIR spectrometry in detecting internal egg abnormalities.
• System configuration, including light source and sensor placement, is crucial for accurate detection.
• Optimized spectral analysis significantly enhances the reliability of abnormal egg identification.