关键词: 可见-近红外光谱, 无损检测, 紫薯, 花青素, 可溶性固形物, 总糖

Abstract: This study aimed to simultaneously, rapidly and nondestructively detect anthocyanins, soluble solid content (SSC) and total sugar (TS) in commercial fresh purple sweet potato by visible and near infrared (VIS/NIR) spectroscopy. The infrared spectra of 52 purple potato cultivars were recorded with the VIS/NIR spectroscopy system set up in our laboratory and Savitzky-Golay (SG) smoothing, standard normal variate (SNV) transformation and first derivative (FD) were separately used to preprocess the original spectra. Partial least squares regression (PLSR) models were established. The best predictive model for anthocyanins and TS was obtained with SG smoothing combined with FD, whereas SNV transformation gave the best predictive model for SSC. Furthermore, new models were developed with the selected optimal pretreatments by competitive adaptive reweighed sampling (CARS) algorithm. For anthocyanins, the correlation coefficient of prediction was 0.942 1, and the root mean square error of prediction was 0.225 9 mg/g; for SSC, the correlation coefficient of prediction was 0.943 1, and the root mean square error of prediction was 0.878 7 °Brix; for TS, the correlation coefficient of prediction was 0.925 3, and the root mean square error of prediction was 0.244 3%. These results showed that using visible/near infrared spectroscopy could achieve the simultaneous, rapid and nondestructive detection of anthocyanins, SSC and TS in fresh purple sweet potato, which is of great practical importance for rapid non-destructive quality testing of fresh purple sweet potato.

Key words: visible/near spectroscopy, nondestructive testing, purple sweet potato, anthocyanins, soluble solids content, total sugar