Method for the Estimation of the Individual Shelf Life of Eggs Using Mapping Scanning with Near-Infrared Spectroscopy

Abstract

　To prevent the loss of eggs used for consumption, it is important to know the individual shelf life of eggs. In this study, a method for the estimation of the individual shelf life of eggs was proposed, based on the diffuse reflectance spectra mapping data obtained from the egg surface. The storage test for the eggs was conducted at a constant temperature, and for the test period, the spectra at 16 points on the surface of each egg, its Haugh Unit (HU) values, and albumen ATP values were measured every 3 days. The HU values were found to be highly correlated to the number of days after laying (DAL) with R²=0957. The relationships between the spectra mapping data and DAL were analyzed using the PLSR method, and ten wavelengths with good correlation performance were nominated. An estimation model for HU and albumen ATP values of eggs based on the spectra and its secondary different data was then constructed using neural networks with 20 input layer neurons, 40 hidden layer neurons, and 2 output layer neurons. The results obtained using the trained neural network estimation model showed good agreement with the measured data, the performance being ±5.5% in average relative error both for HU values and ATP values.