摘要：Visible-near infrared (Vis-NIR) hyperspectral images (400–1051nm) together with chemometrics can be used for the detection of waxed rice. The objective of this study was to find an effective testing method for detecting waxed rice based on the Vis-NIR hyperspectral imaging. Multiplicative scatter correction (MSC) was conducted to preprocess the original spectra. Successive projections algorithm (SPA) was employed for selecting effective wavelengths in the calibration set (200 samples). Based on the effective wavelengths, the predict models were set up using three different models — partial least squares regression (PLSR), multiple linear regression (MLR), and linear discriminant analysis (LDA). Both MSC-SPA-MLR and MSC-SPA-LDA were found to provide 96% detection rate compared to MSC-SPA-PLSR, giving 92% detection rate. Comparative study showed better prediction ability for both MSC-SPA-MLR and MSC-SPA-LDA. Moreover, the hyperspectral imaging technique in the Vis-NIR region could be a reliable method for waxed rice detection.
