Hyperspectral reflectance imaging combined with chemometrics and successive projections algorithm for chilling injury classification in peaches

Chilling injury is one of physiological disorder in peach fruits, which will reduce its' edible and processing quality. In the work, hyperspectral reflectance imaging (400-1000 nm) combined with chemometrics was used to evaluate chilling injury of peaches. Discriminating models including partial least squares-discriminant analysis (PLS-DA), artificial neural networks (ANN), and support vector machines (SVM), were developed for two-class (non-chilled and chilled), three-class (non-chilled, semi chilled and heavy-chilled) and four-class (non-chilled, slight-chilled, moderate-chilled, and heavy-chilled) classifications. The results showed that, using full wavelengths, ANN model had the highest classification rates for the prediction set, with accuracies of 85.37%, 96.11%, and 99.29% for four class, three-class and two-class classifications, respectively. Furthermore, six optimal wavelengths, selected by the successive projections algorithm, were used as the input of PLS-DA, Fisher linear discriminate analysis, ANN and SVM models also presented good performances for two-class classification, with a discriminating accuracies of 92.96%-97.28%. Furthermore, a spatial distribution map of the chilling injury areas was generated by transferring the principal component analysis algorithm of the images. The results showed that the hyperspectral reflectance imaging technique is feasible and useful for the non-destructive detection of peaches' chilling injury, even with several wavelengths, before consumption and processing. (C) 2016 Elsevier Ltd. All rights reserved.