Texture Feature Extraction in Grape Image Classification Using K-Nearest Neighbor

Abstract

Indonesian Grapes are a vine. This fruit is often found in markets, shops, and the roadside. Along with the development of computer technology today, computers can solve problems by classifying objects and objects. How to apply GLCM and K-NN methods for the classification of grapes. The purpose of this study is to apply the GLCM and K-NN methods in the classification of grapes. The dataset used from kaggle.com sources, the data tested are 3 types of grapes, and the number of images is 2624. The fruit that will be used for the data collection and classification process is limited to three types of grapes, namely grape blue, grape pink, and grape white. How to apply GLCM and K-NN methods for the classification of grapes. The feature extraction of GLCM used in this study is the feature contrast, energy, correlation, and homogeneity. From testing the test data, the highest accuracy value is 99.5441% with k = 2 at level 8, while the lowest accuracy value is 24.924% at each k level 2. The GLCM level value is very influential on the accuracy results, namely, the higher the GLCM level value, the higher the GLCM value. accuracy is getting better.