Intelligent Post Harvest Quality Prediction of Fruits Using Non-Destructive Spectral Analysis
Journal of Smart and Sustainable Farming · 18.07.2026 · 150
DOI: 10.64026/jssf/2026013 ↗Kurz gefasst
Fachartikel in der Zeitschrift Journal of Smart and Sustainable Farming, online erschienen am 18.07.2026. Verfasst von Lise Roy Porter und 1 weiteren Autor:innen. Die Arbeit nennt 17 Literaturangaben.
Abstract (Originalsprache)
Post-harvest losses of fruit are mainly caused by the absence of external quality as an indicator of internal quality (ripeness, eating quality and remaining shelf life) which are parameters that cannot be assessed from external inspection, and the conventional methods used to assess these parameters (soluble solids content, SSC and firmness) are destructive, slow and require only small portions of fruit. An intelligent framework has been developed to predict the internal quality and storage stage of mango using the diffuse reflectance spectra (DRS) (600–1700 nm) of the fruit in the visible–near-infrared (Vis–NIR) region, obtained through diffuse reflectance spectroscopy (DRS), and hyperspectral images. The framework integrates absorbance conversion, scatter correction (standard normal variate, SNV), Savitzky–Golay (SG) derivatives, partial least squares regression (PLSR), iterative wavelength elimination and machine-learning regressors with the extension to point predictions to pixel-wise quality maps and first-order shelf-life kinetics. Within the set of 240 fruits stored for 0–14 days (160 were used for calibration and 80 were used for independent prediction), the PLSR model with SNV + SG first-derivative spectra predicted SSC with R2P = 0.91, RMSEP = 0.93 °Brix and RPD = 3.3, while firmness was predicted with R2P = 0.92, RMSEP = 4.6 N and RPD = 3.6. The accuracy of the firmness remained similar when 36 of 221 wavelengths were selected, and this accuracy was comparable to that for a support vector regressor, and inferior to that for a multilayer perceptron, with the available sample size. The three maturity stages were classified correctly with 75 % accuracy and the maturity stage at which firmness dropped to 15 N was predicted with 8.2 d against 8.1 d measured. The method allows for fast grading, control of ripening room conditions and logistics decisions based on the shelf life without harming the fruit.
Zitieren
APA
Porter, L. R., & Cherinet, M. S. (2026). Intelligent Post Harvest Quality Prediction of Fruits Using Non-Destructive Spectral Analysis. Journal of Smart and Sustainable Farming, 150. https://doi.org/10.64026/jssf/2026013
Vancouver
Porter LR, Cherinet MS. Intelligent Post Harvest Quality Prediction of Fruits Using Non-Destructive Spectral Analysis. Journal of Smart and Sustainable Farming. 2026:150. doi:10.64026/jssf/2026013
BibTeX
@article{Porter2026,
author = {Porter, Lise Roy and Cherinet, Mekdes Solomon},
title = {Intelligent Post Harvest Quality Prediction of Fruits Using Non-Destructive Spectral Analysis},
journal = {Journal of Smart and Sustainable Farming},
year = {2026},
pages = {150},
doi = {10.64026/jssf/2026013}
}
EUVIMED