Abstract
This study tested a non-destructive technique for predicting quality indices of SO(2) pre-treated and dehydrated mangoes. Near-infrared hyperspectral imaging (NIR-HSI), a non-destructive technique, was tested for classifying and predicting total soluble solids (TSS) and sulfur dioxide (SO(2)) content of the treated mangoes. The samples were scanned through a laboratory-based near infrared hyperspectral imaging system within the wavelength range of 935-1720 nm in order to acquire the spectral data. The calibration models were developed for quality indices (dependent variables) by spectral data (independent variables) using partial least square regression (PLSR). Savitzky- Golay smoothing pretreatment was used for creating the calibration model for TSS while the original spectra were used for creating the calibration model for SO(2) content. The models obtained predictive results for TSS and SO(2) content with correlation coefficient of prediction (R(p)) 0.82 and 0.83 respectively and root mean square error of prediction (RMSEP) of 2.42% and 56.40 mg/kg, respectively. The models were used to predict TSS and SO(2) content in each pixel of each SO(2) pre-treated and dehydrated mango image using HSI scanning and interpolated to a linear-color-scale in order to obtain the predictive images, which showed TSS and SO(2) content of SO(2) pre-treated and dehydrated mango by color visualization. The results showed that NIR-HSI could possibly be used to predict TSS and SO(2) content. These factors are important quality indices of SO(2) pre-treated and dehydrated mangoes. SUPPLEMENTARY INFORMATION: The online version contains supplementary material available at 10.1007/s13197-024-06132-8.