An efficient ANFIS based pre-harvest ripeness estimation technique for fruits

Abstract

The ripeness estimation of fruits plays a significant role in marketing and evaluation of quality. However, due to the subjectivity and slow speed in the case of manual assessment, the agriculture industry leads to the need for automation. In this research work, an efficient ANFIS based Pre-harvest Ripeness Estimation (APRE) technique is proposed for the ripeness estimation of fruits based on color. There are three main phases of the proposed work: Data Processing, Feature Selection and Fuzzy Logic Implementation. In the first phase, the data set of images of fruits is prepared in the image acquisition phase. Then images are pre-processed to make them equal in size. In Image Segmentation phase, a fruit is extracted from its background. In the next phase, two color features: red-green color difference and red-green color ratio are calculated based on extracted RGB color attributes and R-G is chosen based on performance comparison in terms of classification accuracy. In phase three, Adaptive Neuro-Fuzzy Inference System (ANFIS) is utilized for designing and implementing the classification system which classifies the fruits into six ripeness phases. The experimental results show that the APRE performs better than SVM, Decision Tree, and KNN in terms of accuracy, precision, recall, sensitivity and F-measure.