Classification of Troso Fabric Using SVM-RBF Multi-class Method with GLCM and PCA Feature Extraction

Abstract

Image Processing has many benefits that we have seen, such utilization is one of them to help analyze images by classification, image pattern recognition, data security, and copyright protection. Troso fabrics are classified using a combination of two methods. Using the Gray Level Co-occurrence Matrix (GLCM) and Principle Component Analysis (PCA) method for feature extraction and multiclass Support Vector Machine (SVM) used is Ones Against All (OAA) & Ones Against One (OAO) with the type of gaussian kernel or Radial Basis Function (RBF) as a classification method. Testing of the two methods was carried out on 3 types of troso fabric using images measuring 480x480 pixels. From the results of tests that have been carried out, the classification of troso fabric produces an accuracy of 90% for SVM OAA and 86,7% for SVM OAO for the best method feature extraction using GLCM.