Evaluating ripeness in post-harvest stored kiwifruit using VIS-NIR hyperspectral imaging

Kiwifruit ( Actinidia deliciosa ) stored long-term at low temperatures after harvest can exhibit varying internal quality upon shipment due to the influence of harvest conditions. Flesh firmness (FF) and soluble solids content (SSC) are attributes of eating quality and ripeness, which change during storage. To ensure the timely shipment of kiwifruit with uniform quality, the development of non-destructive measurement techniques for FF, SSC, and ripeness is necessary. In this study, models were developed to predict the FF, SSC, and ripeness stages of kiwifruits using visible-near-infrared (Vis-NIR) hyperspectral imaging. The FF and SSC of kiwifruit were investigated according to the storage period, and five ripeness stages were defined based on these characteristics. Vis-NIR hyperspectral images of kiwifruit stored for 0–120 d were measured to extract hyperspectral spectra. Partial least squares regression and support vector machine regression (SVMR) prediction models were developed to predict the FF and SSC of kiwifruit, and partial least square-discriminant analysis (PLS-DA) and support vector machine classification (SVMC) models were developed to classify ripeness stages. The SVMR model with second-order derivative preprocessing exhibited the best performance in FF prediction, with the results of R 2 p and root mean square of prediction (RMSEP) and RDP values as 0.878, 3.008 N and 2.721, respectively. For SSC prediction, the SVMR model with multiplicative scatter correction (MSC) preprocessing exhibited the best performance, with the results of R 2 p and RMSEP and RPD values as 0.940, 0.898 °Brix and 4.055, respectively. Ripeness determination achieved the highest accuracy of 91.463 % and 91.548 % for the PLS-DA and SVMC models with maximum normalization preprocessing and range normalization preprocessing, respectively. The results of this study demonstrate that Vis-NIR hyperspectral imaging is useful for rapidly identifying the internal quality and post-storage ripeness stages of kiwifruit stored at low (0℃) temperatures. Furthermore, the developed technology is expected to contribute to determining the optimal shipping time for kiwifruit during storage. • FF and SSC of kiwifruit were investigated according to the storage period (0–120 d). • FF and SSC-based kiwifruit ripeness stage criteria were presented to determine the optimal shipping time. • Vis-NIR hyperspectral imaging was utilized to accurately assess kiwifruit quality and ripeness. • PLS and SVM models were developed for quality prediction and ripeness classification of long-term stored kiwifruit. • The SVM model achieved the best internal quality prediction performance and ripeness classification performance.