Estimating bulk optical properties of AFB1 contaminated edible oils in 300–900 nm by combining double integrating spheres technique with laser induced fluorescence spectroscopy

An optical detection platform based on laser induced spectroscopy and double integrating spheres techniques was developed to obtain absorption ( μ a ), reduced scattering coefficients ( μ' s ) and fluorescence intensity of oil. The validation experiment carried on liquid phantoms showed that the developed system could achieve high linearity, and the results of spectra analysis indicated that the fluorescence intensity has a significant negative correlation with both μ a and μ' s . A total of 1620 oils with six categories were detected. The reason for the difference of fluorescence and μ a spectra was analyzed by comparing the measured chlorophyll, polyphenol and α-tocopherol contents. Linear discriminant analysis combined with principal component analysis based on fluorescence and μ a spectra was employed, to calibrate the AFB 1 classification models. The discrimination results manifested that by integrating μ a with fluorescence signal, the correct classification rate could be improved by more than 10%, and the false negative rate was greatly reduced.