Towards importance of comprehensive color features analysis using iterative golden ratio proportions for Alphonso mango ripening stage classification by adapting to natural progressive ripening method

The color and texture of the mango fruit are the primary qualitative factors that play a crucial role during fruit sampling for production. For mango fruit industrial processing, we consider the batches of mango fruit to interpret its texture and color properties to determine its ripening stage into unripe, partial ripe, and ripe. To understand the ripening stage of mango fruit, the localized features are computed from the two-level decomposition of fruit using the golden ratio. The features computed from decomposed segments are analyzed using color features to classify the fruit into a specific stage using classifiers, support vector machine, random forest, K nearest neighbor, gradient boost, Adaboost, and decision trees. For experimental analysis, the dataset of 2400 samples of Alphonso mangoes was considered for feature extraction using mean color features from five color spaces and classified with an accuracy of 97.6% with the SVM and Random Forests. The analysis is also conducted with features extracted from the whole fruit and achieved an accuracy of 77.39% with support vector machines using polynomial kernel and random forests. Extended analysis is also done on color features extracted separately from each color space and with specific decomposed slice combinations. From the outcomes achieved, it is evident that the proposed approach achieves promising outcomes and a steep improvement in accuracy over the state-of-the-art models.