Comparison of a dual-laser and a Vis-NIR spectroscopy system for detection of chilling injury in kiwifruit

•A dual laser system was developed for the detection of chilling injury in kiwifruit.•Supervised classification was made with a support vector machine algorithm.•Classification performance of the dual laser system was compared with Vis-NIR spectroscopy.•Two-class and three-class classifications were...

Ausführliche Beschreibung

Gespeichert in:
Bibliographische Detailangaben
Veröffentlicht in:Postharvest biology and technology 2021-05, Vol.175, p.111418, Article 111418
Hauptverfasser: Wang, Zhen, Künnemeyer, Rainer, McGlone, Andrew, Sun, Jason, Burdon, Jeremy
Format: Artikel
Sprache:eng
Schlagworte:
Online-Zugang:Volltext
Tags: Tag hinzufügen
Keine Tags, Fügen Sie den ersten Tag hinzu!
Beschreibung
Zusammenfassung:•A dual laser system was developed for the detection of chilling injury in kiwifruit.•Supervised classification was made with a support vector machine algorithm.•Classification performance of the dual laser system was compared with Vis-NIR spectroscopy.•Two-class and three-class classifications were studied for both systems.•Wavelength sensitivity to kiwifruit chilling injury was explored. A novel dual-laser system with laser wavelengths of 730 nm and 850 nm has been developed for the non-destructive detection of chilling injury in Actinidia chinensis var. chinensis ‘Zesy002’ kiwifruit. Chilling injury in kiwifruit is a physiological disorder that may occur during low-temperature storage, with symptoms that are often not evident until the fruit is cut open. This study involved a sample of 162 kiwifruit with differing severity of chilling injury, and a performance comparison between the novel dual-laser system and a standard visible – near infrared (Vis-NIR) interactance spectroscopy approach proven in a prior study. The dual-laser system involved scanning the laser beams across the fruit to generate spatial profiles of light transmission in the fruit. Data analysis with supervised model training, using a support vector machine algorithm, was successfully used to achieve cross-validation prediction accuracies higher than 90 % for distinguishing sound and chilling-injured kiwifruit. The performance was equivalent to that achieved by the Vis-NIR interactance spectroscopy approach, suggesting that the dual-laser method is an alternative and more attractive option because of its easier system layout for high-speed on-line sorting of kiwifruit for chilling injury disorder.
ISSN:0925-5214
1873-2356
DOI:10.1016/j.postharvbio.2020.111418