Potential of multispectral imaging for real-time determination of colour change and moisture distribution in carrot slices during hot air dehydration

Colour and moisture content are important indices in quality monitoring of dehydrating carrot slices during dehydration process. This study investigated the potential of using multispectral imaging for real-time and non-destructive determination of colour change and moisture distribution during the hot air dehydration of carrot slices. Multispectral reflectance images, ranging from 405 to 970 nm, were acquired and then calibrated based on three chemometrics models of partial least squares (PLS), least squares-support vector machines (LS-SVM), and back propagation neural network (BPNN), respectively. Compared with PLS and LS-SVM, BPNN considerably improved the prediction performance with coefficient of determination in prediction (RP2) = 0.991, root-mean-square error of prediction (RMSEP) = 1.482% and residual predictive deviation (RPD) = 11.378 for moisture content. It was concluded that multispectral imaging has an excellent potential for rapid, non-destructive and simultaneous determination of colour change and moisture distribution of carrot slices during dehydration.