IRIS-GRAPE: an approach for prediction of quality attributes in vineyard grapes inspired by iris biometric recognition

Abstract

The determination of the period of harvest is essential in the vineyard. For that, attributes of quality such as Total Soluble Solids (TSS) and phenolic compounds are constantly monitored along the maturation. This work proposes a new non-destructive approach for prediction of TSS, total anthocyanins and yellow flavonoids using RGB images, called IRIS-GRAPE. It is inspired by the process of biometric recognition using the iris. In order to validate the proposed approach, a study comparing its performance with the traditional method was performed, using the average of RGB pixel values of the region of interest (ROI) as input variables for a multiple linear regression. The study used two performance evaluation metrics: the correlation coefficient (rP) and the mean square error (MSE). In order to compare the performance differences between the proposed approach, IRIS-GRAPE, and the traditional approach, hypothesis tests were done. The results show that the proposed approach has a superior performance than the traditional method with a confidence level of 95%