Klasifikasi Jenis Pisang Menggunakan Support Vector Machine dengan Fitur GLCM dan HOG

Abstract

Banana fruit supplies not only the domestic market, but also the international market. In the process of introducing a variety of bananas are generally done in two ways, firstly done manually by humans to introduce bananas and secondly to use destructive methods by taking samples. The problem that occurs in this second process is having relatively large and greater costs. This requires a system that can classify bananas using digital image processing and the Vector Support Machine (SVM) implemented in this study. The image of a banana is taken with a Xiaomi Note 4x cellphone camera and is processed using Matlab software. Digital images are used to extract the shape and texture features of bananas, while SVM is used for banana classification. This study uses 420 images of bananas divided into 7 classes, namely Ambon banana class, Barangan banana class, golden banana class, Kepok banana class, Raja banana class, milk banana class and banana horn class. Where in the test using cross validation for 7 classes of bananas. SVL is able to classify the types of bananas in the image with GLCM and HOG features in iteration 1 with an overall accuracy of 74.28% in the type of milk bananas