10:30 AM - 10:45 AM
[6-1015-C-02] Use of hyperspectral imaging to separate cultivars and evaluate the internal quality of nectarines
Keywords:Stone fruit, Hyperspectral transmittance imaging, Ripeness monitoring, Cultivar discrimination, Internal quality, Computer vision
Visible–near-infrared hyperspectral imaging (450-1040 nm) was studied in reflectance and transmittance modes to assess the internal physicochemical properties and sensory perception of ‘Big Top’ and ‘Magique’ nectarines (Prunus persica L. Batsch var. nucipersica) (yellow and white-flesh cultivar, respectively) during ripening. These properties were successfully correlated to the Ripening Index (RPI) and the Internal Quality Index (IQI). During ripeness under controlled conditions, hyperspectral images of the intact fruits were acquired and their physicochemical properties (flesh firmness, total soluble solids, titratable acidity and flesh colour) were analysed. Moreover, a model to discriminate between both cultivars were developed. IQI and RPI were correlated to the hyperspectral images by using Partial Least Square (PLS) regression with proper variables selection. Optimal results were obtained with R2 (and RPD) values of 0.89 (2.7), 0.90 (3.1), 0.90 (2.8) and 0.88 (2.7) for RPI and IQI in ‘Big Top’ and ‘Magique’ nectarines, respectively.
In addition, the emergence of new cultivars in the market with similar appearance but different sensory properties can cause confusion among the consumers, being necessary the development of new tools capable of discriminating these cultivars in an automated and non-destructive way. PLS-DA was used to obtain the best classification model to distinguish intact fruits of both cultivars using individual pixel spectrum and mean spectrum of each fruit, and then projecting the model onto the complete surface of fruits in a validation or prediction set. The results indicated that mean spectrum approach was the most accurate, 84.4 % vs. 94.4 %. Moreover, a comprehensive wavelength selection was performed, reducing the dimensionality of the hyperspectral images using the regression coefficients of the mean spectrum PLS-DA model, obtaining an accuracy of 96.3 % by using 14 optimal wavelengths.
A PLS model of IQI prediction was used to transfer the calibrated results to each pixel of the image and to visualise the evolution of ripeness on the surface of the fruits, and also to represent the probability of whether any pixels belongs to one or another cultivar.
Finally, the internal quality of the nectarines was inspected using hyperspectral transmittance imaging during their ripening under controlled conditions. The detection of split pit disorder and classification according to an established firmness threshold were performed using PLS-DA. The prediction of the IQI related to ripeness was performed using PLS-R. The most important variables were selected using interval-PLS. As a result, an accuracy of 94.7 % was obtained in the detection of fruits with split pit of the ‘Big Top’ cultivar. Accuracies of 95.7 % and 94.6 % were achieved in the classification of the ‘Big Top’ and ‘Magique’ cultivars, respectively, according to the firmness threshold. The internal quality was predicted through the IQI with R2 values of 0.88 and 0.86 for the two cultivars. The results obtained indicate the great potential of hyperspectral
In addition, the emergence of new cultivars in the market with similar appearance but different sensory properties can cause confusion among the consumers, being necessary the development of new tools capable of discriminating these cultivars in an automated and non-destructive way. PLS-DA was used to obtain the best classification model to distinguish intact fruits of both cultivars using individual pixel spectrum and mean spectrum of each fruit, and then projecting the model onto the complete surface of fruits in a validation or prediction set. The results indicated that mean spectrum approach was the most accurate, 84.4 % vs. 94.4 %. Moreover, a comprehensive wavelength selection was performed, reducing the dimensionality of the hyperspectral images using the regression coefficients of the mean spectrum PLS-DA model, obtaining an accuracy of 96.3 % by using 14 optimal wavelengths.
A PLS model of IQI prediction was used to transfer the calibrated results to each pixel of the image and to visualise the evolution of ripeness on the surface of the fruits, and also to represent the probability of whether any pixels belongs to one or another cultivar.
Finally, the internal quality of the nectarines was inspected using hyperspectral transmittance imaging during their ripening under controlled conditions. The detection of split pit disorder and classification according to an established firmness threshold were performed using PLS-DA. The prediction of the IQI related to ripeness was performed using PLS-R. The most important variables were selected using interval-PLS. As a result, an accuracy of 94.7 % was obtained in the detection of fruits with split pit of the ‘Big Top’ cultivar. Accuracies of 95.7 % and 94.6 % were achieved in the classification of the ‘Big Top’ and ‘Magique’ cultivars, respectively, according to the firmness threshold. The internal quality was predicted through the IQI with R2 values of 0.88 and 0.86 for the two cultivars. The results obtained indicate the great potential of hyperspectral