Detection Of Eggshell Strength And Thickness Based On NIR Spectra

Eggshell quality is one of the important characteristics of quality of eggs. It’s very important for the hatch, storage, transportation and sale of eggs. But in the turnover process from production to sale, egg breakage is a common problem, and the hazard is not only losing their edibleness, but also causing more serious damage by contaminating other eggs and leading to cross-contamination. So, eggshell quality detection is essential.The research object is jinqi eggs. Detection of eggshell quality is studied with near-infrared (NIR) spectroscopy, Chemometrics analysis method and Physical index analysis. The main contents and conclusions are as follows:(1)The NIR spectra of eggs and conventional eggshell quality indicators were collected; the correlation between conventional eggshell quality indicators was studied. The results showed that the correlation is big between eggshell strength with eggshell thickness, eggshell percentage and egg specific gravity, the correlation coefficient were0.79,0.76and0.62, respectively. While the correlation is small between eggshell strength with egg shape index, the correlation coefficient is only-0.14, which was negatively correlated; that means, the greater the eggshell strength, the more round the egg is.(2)The eggshell ultra-microstructures of three groups samples with high, medium, and low eggshell strength were compared. The results showed that with the increase of the eggshell strength, the eggshell cross-sectional consistency and smoothness became better, the mastoid layer and fiber layer related more closely and the effective thickness of the mastoid became smaller, the outer surface of eggshell became smoother and denser, and cracks reduced; the eggshell fiber layer structure became closer, the primary and secondary branch was clearer, fiber gap became smaller.(3)The effects of eggshell surface structure on the NIR spectral characteristics were studied. The results showed that the average spectral reflectance of the high-strength group was biggest, that of the medium-strength group is in the second place, and that of the low-strength group is smallest.(4)The quantitative models were established for eggshell strength and eggshell thickness after the abnormal samples excluded by Chauvenet test and Leverage and students residual T test, The results showed that:the optimal performance model of eggshell strength was partial least squares regression (PLSR) model within full band range with multiplicative signal correction (MSC) pretreatment, the correlation coefficients of the calibration and prediction sets were0.912and0.748, the root mean squre error of calibration (RMSEC) and root mean square error of prediction (RMSEP) was2.24and3.53N, respectively. Meanwhile, correlation coefficients of the calibration and prediction sets for eggshell thickness were0.931and0.86, RMSEC and RMSEP was0.0099and0.0139mm. Parameter quantitative compensation models using egg shape index, egg weight, eggshell thickness, eggshell percentage, egg specific gravity and multi-parameter of them for eggshell strength were established, the results showed that all of compensation models performance were improved, and multi-parameter compensation model was the best.(5)According to the distribution of the eggshell strength and eggshell thickness, the egg samples were divided into three groups of high, medium and low eggshell strength and eggshell thickness, then, the qualitative models were established.The results showed that:the qualitative discriminant optimal model of eggshell strength is discriminant PLS (DPLS) model within full band range with standard normal variate (SNV) pretreatment, the correct rate of the calibration set and prediction set were88%and79.59%, respectively. The qualitative discriminant optimal model of eggshell thickness was discriminant analysis (DA) model within4000-7500cm"’band range using13principal components combined with SNV and35or45points smoothing preprocessing, the correct rate of the calibration set and prediction set were81%and85.71%, respectively. Parameter qualitative compensation models using egg shape index, egg weight, eggshell thickness, eggshell percentage, egg specific gravity and multi-parameter of them for eggshell strength were established, the poor performance of egg shape index and egg weight compensation model may result from the small correlation between them with eggshell strength, which interfered the discrimination accuracy of the model.The discrimination results of the remaining compensation models had been improved, and the result of the multi-parameter compensation model was the best.(6)The results of quantitative and qualitative analysis of eggshell strength and eggshell thickness showed that the detection of eggshell quality using NIR spectroscopy was feasible, but the accuracy remained to be improved.