Computer Vision System Applied to Classification of Medlar (Mespilus germanica) during ripening stage at cold storage

Today, computer vision system (CVS) has become an alternative to visual inspection being objective, consistent, rapid, and economical in various agricultural and food industry commodity grading systems,. A particular application of this technology is the estimation of ripening or the study of the evolution of maturity of several produce in order to improve storage conditions or to be able to offer consumers better products. Short storage life of medlar fruit (Mespilus germanica) and its high susceptibility to water loss and browning are the main factors limiting its marketability. The aim of this work was to implement a straightforward and low-cost method at laboratory scale as an initial approach, in order to determine the ripening stages of M. germanica by means of a CVS and multivariate analysis. In the present work, physicochemical properties and color parameters obtained using a CVS at laboratory level were linked to establish the ripening stages of M. germanica. To classify the stages, a ripening index (RPI) was proposed, in which three stages were identified; unripe, ripe and senescent. Two classifiers based on principle component analysis (PCA) and multivariate discriminant analysis (MDA) were used to assess the applicability of vision system. The color parameters correlate correctly with the physicochemical changes which are considered the standard method to evaluate the maturity of fruits. PCA made it possible to obtain classification rates of 92.11% and 95.31% with and without physicochemical parameters, respectively. MDA was capable of classifying apples in their correct ripening stage with 96.08% accuracy. The results obtained showed that CVS developed for the study can be used as a useful non-invasive, efficient method for the evaluation of the ripeness of mangoes.