Detection of Astringent and Deastringent Persimmon Fruits using Hyperspectral Imaging Technology
Derechos de accesoopenAccess
MetadataShow full item record
AuthorMunera, Sandra; Aleixos, Nuria; Gómez-Sanchís, Juan; Besada, Cristina; Cubero, Sergio; Talens, Pau; Salvador, Alejandra; Blasco, José
Cita bibliográficaMunera, S., Aleixos, N., Gómez-Sanchís, J., Besada, C., Cubero, S., Talens, P. et al. (2018). Detection of Astringent and Deastringent Persimmon Fruits using Hyperspectral Imaging Technology. Proceedings of the European Conference on Agricultural Engineering (AgEng2018), 946-950.
Persimmon fruit cv. ‘Rojo Brillante’ is an astringent cultivar due to its content of soluble tannins. Traditionally, the consumption of this cultivar has been only possible when the astringency has been naturally removed before harvest, when fruit is overripe and the manipulation is very delicate. In recent years, new postharvest treatments, which allow astringency removal while preserving high flesh firmness, have been developed. Among them, the most widely used in commercial settings is based on exposing fruits to high CO2 concentrations for 24 h–36 h. This method promotes anaerobic respiration in the fruit, giving rise to an accumulation of acetaldehyde and insolubilizing tannins at the end of the treatment. The effectiveness of this treatment is controlled by means of methods that are destructive, time-consuming and only a few samples per batch can be analysed. For this reason, the objective of this work is to study the application of the hyperspectral imaging technology in the detection of astringent and deastringent fruits non-destructively. A total of 300 fruits were used and exposed to CO2 during different times in order to obtain fruit with different content of soluble tannins. The hyperspectral images of the fruits were acquired using a VIS-NIR hyperspectral system, which covers the spectral range 450-1040 nm. A reference analysis of soluble tannins was performed in order to find out if the fruits were astringent or deastringent. The spectral information of the two thirds of the fruits was used to build the classification models by means of partial least squares (PLS) and support vector machine (SVM) discriminant analysis methods. The remaining third was used to validate the models as test set. As result, 92.6 % astringent and 84.4 % deastringent fruits were classified correctly using the SVM method. This shows the great potential of hyperspectral imaging technology to detect astringent and deastringent fruits in industrial setups.