VISUAL AND NEAR-INFRARED REFLECTANCE SPECTROSCOPY FOR RICE TASTE EVALUATION

Abstract

Visible and near-infrared (VIS/NIR) spectroscopy calibration models for rice taste evaluation were developedusing 61 short-grain rice samples. The best performance calibration model was obtained from original spectra of wholegrain milled rice using multiple linear regression (MLR) analysis. The correlation coefficient (r) and the standard error ofprediction (SEP) of the validation set was 0.62 and 0.27, respectively. However, this was not adequate to justify replacingsensory tests with the calibration model for evaluating rice taste. The results indicated that VIS/NIR technology could beused for classifying rice samples into qualitative groups, such as poor taste, better taste and the best taste.