Application of hyperspectral imaging for prediction of physico-chemical and sensory characteristics of table grapes

The possibility of applying the hyperspectral imaging technique for prediction of some physico-chemical and sensory indices of table grapes was checked. Seven cultivars were studied: Italia, Baresana, Pizzutello, Red Globe, Michele Palieri, Crimson Seedless, and Thompson Seedless. A hyperspectral imaging system was used to acquire the reflectance spectra of berries. Successively, the same berries were analysed for their pH, total acidity, and soluble solid content according to common methods. Quantitative descriptive sensory analysis was performed by a trained panel. A Partial Least Squares Regression (PLSR) model was applied in order to find correlations between spectra information and each of the physico-chemical indices. Good correlations were found between each of the physico-chemical indices and the spectra information. Concerning titratable acidity, coefficients of determination were equal to 0.95 and 0.82 for white and red/black grapes, respectively whereas the relative values for soluble solid content were 0.94 and 0.93, and for pH 0.80 and 0.90. Spectra information was not correlated with the sensory data, making hard prediction of attribute perception.

► The application of the hyperspectral imaging to prediction of grape quality was checked. ► Four white grapes and three red grapes were studied. ► It was possible to predict soluble solid content, acidity, and pH. ► The absence of correlation between spectra and sensory data was highlighted.