Abstract:
In order to sort good qualities of guava, titratable acidity (TA), total soluble solids (TSS) and firmness of guavas as important indices that influence the acceptance of consumers were considered. Therefore reflectance near infrared hyperspectral imaging (NIR-HSI) in the wavelength of 900-1700 nm which is a non-destructive technique was used to predict TA, TSS and firmness of guavas in this study. Guavas were scanned by NIR-HIS and divided into 2 groups for calibration and prediction. Average spectrum from a region of interest (ROI) of each sample from NIR-HSI was used to create calibration models to predict TA, TSS and firmness. The calibration models were established using the partial least squares regression (PLSR). The prediction results obtained the coefficient of determination (R2) of 0.972 and the root mean squared error of prediction (RMSEP) of 0.010% for TA, R2 of 0.801, RMSEP of 0.437 oBrix for TSS and R2 of 0.747, RMSEP of 0.610 N for firmness. Hence, the near infared hyperspectral imaging (NIR-HSI) technique was feasible and had the potential to be used to measure the qualities of guavas.