On-line detection of blood spot introduced into brown-shell eggs using visible absorbance spectroscopy

Blood spots in eggs affect their quality, but it is difficult to realise on-line detection of blood spots in brown-shell eggs because the absorption feature of pigment in brown eggshell is similar to that of the blood spot. The major purpose of this study is to explore the optimal discrimination method based on visible absorbance spectroscopy to realise on-line detection. The spectra of 96 brown-shell normal eggs and 98 brown-shell artificial blood-spot eggs were collected in the spectral range of 200-1100 nm by a prototype egg internal quality detection system with a conveyor speed of 4 eggs per second. Three discrimination methods, partial least squares discriminant analysis (PLS-DA), k-nearest neighbour (KNN) and binary logistic regression (BLR) were used and compared. The results showed that the BLR method was better than PLS-DA and KNN, and the best discrimination rates for the training set and prediction set were 95.4% and 96.9%, respectively. In an external validation with 220 eggs, all three real blood-spot eggs were detected and no normal egg was misjudged by the egg internal quality detection system with BLR model. These results indicated that visible absorbance spectroscopy combined with BLR model could be applied as an on-line detection tool for brown-shell blood-spot eggs.