An intelligent system for egg quality classification based on visible-infrared transmittance spectroscopy

The potential of the visible infrared (Vis–IR) (400–1100 nm) transmittance method to assess the internal quality (freshness) of intact chicken egg during storage at a temperature of 30 ± 7 °C and 25 ± 4% relative humidity was investigated. Two hundred chicken egg samples were used for measuring freshness and spectra collection during egg storage (up to 25 days). Two correlation models, firstly between Haugh unit (HU) and storage time, and secondly between the yolk coefficient (YC) and storage time, were developed and yielded correlation coefficients (R2) of 0.86 and 0.96, respectively. These models spanned the period for which egg quality decreased dramatically and are statistically significant (P < 0.05). In addition, to reduce the dimensionality of the spectra and extract effective wavelengths, two methods were developed based on principal component analysis (PCA) and a genetic algorithm (GA). The output of PCA and GA were also used comparatively to design an egg quality intelligent system. The result of the analyses indicated that identification ratio of GA with fast Fourier transform (FFT) preprocessing was superior to other methods, and that the quality classification rates of this method for one-day-old eggs are 100%. This study shows that identification of an egg’s freshness using NIR spectroscopy with GA and artificial neural network (ANN) is reliable.