Quality assessment and discrimination of intact white and red grapes from Vitis vinifera L. at five ripening stages by visible and near-infrared spectroscopy

Abstract

Abstract The diffuse reflection visible/near-infrared (Vis/NIR, 400–1100 nm) and near-infrared (NIR, 900–2500 nm) spectrum were used to monitor the surface color (CIE L*a*b*), total soluble solid contents (SSC) and total phenolic compounds (TP) of intact ‘Manicure Finger’ and ‘Ugni Blanc’ berries at five ripening stages (i.e., green, pre-veraison, veraison, post-veraison and ripe). The determination of quality parameters and the discrimination of five ripening stages were conducted by chemometric analysis based on full-band and selected wavelengths of Vis/NIR and NIR. The results showed that the best regression results were obtained by least squares support vector machine (LS-SVM) with the root mean squares error of prediction (RMSEP) of 5.161, 2.919, 3.275, 1.230% and 0.216 g kg−1 for L*, a*, b*, SSC and TP of ‘Manicure Finger’ in the range of 400–1100 nm, respectively; and the RMSEP of 3.049, 0.710, 2.996 and 0.150 g kg−1 for L*, a*, b* and TP of ‘Ugni Blanc’ in the range of 400–1100 nm, respectively, and the RMSEP of 1.288% for SSC in the range of 900–2500 nm. A total of 90% and 100% classification accuracies on prediction sets were reached by the total soluble solid contents based competitive adaptive reweighted sampling support vector machine discrimination analysis (SSC-based CARS SVM-DA) for ‘Manicure Finger’ and ‘Ugni Blanc’ grape berries of five ripening stages, respectively. This study provided a feasible evaluation method of quality and developing stages for grape varieties during ripening stages by Vis/NIR and NIR technology.