The Classification Of Monster And Williams Pear Varieties Using K-Means Clustering And K-Nearest Neighbor (KNN) Algorithm

Abstract

Abstract Pear is a kind of fruits which has a lot of varieties. One of the way to differ pear varieties is by looking at the color, size and shape. This research is aimed at giving assistance to classify two varieties of pear i.e. Monster Pear and Williams Pear. In order to get the purpose, pear image processing is done to ease the classification process of pear varieties. Research method used consists of RGB Color Room to l*a*b, image segmentation, characteristics extraction with K-Means Clustering. Besides, K-Nearest Neighbor (KNN) is used to know the distribution and the classification. The use of practice data will increase the accuracy of pear varieties classification. The data used here are 88 pear images which cover 44 image practice data of Monster pear and 44 image practice data of Williams pear. Meanwhile, the test data taken are 10 images of each variety. The result of this research shows that the pear classification accuracy level is 95% which is very good.