In-field and non-invasive determination of internal quality and ripeness stages of Feicheng peach using a portable hyperspectral imager

Abstract

In-field and non-invasive determination of internal quality and ripeness stages allows for a selective harvest for Feicheng peach. In this study, a portable hyperspectral imager was used for on-site capturing the images of mid-ripe and ripe peaches on trees, and soluble solids content (SSC) and firmness of the fruit were measured as the reference standards. These samples were split into calibration set and validation set by samples set partitioning based on joint X–Y distances (SPXY) algorithm. Multiple linear regression (MLR) models were established using effective wavelengths selected by competitive adaptive reweighted sampling (CARS) and random frog (RF), respectively. The more promising performances were achieved by RF-MLR model with R v 2 of 0.88 and RMSEV of 0.54 for SSC, and CARS-MLR model with R v 2 of 0.81 and RMSEV of 1.17 for firmness. Furthermore, LIBSVM model was employed to discriminate the ripeness of Feicheng peach using effective wavelengths selected by sequential forward selection (SFS) algorithm, with classification accuracy of 91.7% in the validation set. It can be concluded that portable hyperspectral imager can be applied to determine the internal quality and ripeness stages of Feicheng peach in orchard, providing support for the in-field and nondestructive quality inspection and timely harvesting of Feicheng peach. • In-field capture of hyperspectral images of mid-ripe and ripe Feicheng peaches. • Establish the predictive models for soluble solids content and firmness. • Develop the discrimination model of Feicheng peaches ripeness stages.