Classification of Hijab Types Based on Gray Level Co-occurrence Matrix Features and the K-Nearest Neighbor (KNN) Algorithm

Abstract

This study aims to build an automatic classification system to address the challenge of visually identifying hijab types by utilizing digital image processing technology. The research scope is limited to two categories: pashmina and instant hijabs. The applied method involves the Gray Level Co-occurrence Matrix (GLCM) to extract texture features in four angular directions, which yields four primary feature values: Contrast, Energy, Correlation, and Homogeneity. These features are subsequently classified using the K-Nearest Neighbor (KNN) algorithm with the Euclidean Distance metric. The dataset used consists of 60 image samples, divided into 48 training data and 12 test data. Testing was conducted with varying K-values (1, 3, 5, and 7). The results show that the classification system using the GLCM and KNN combination is effective, achieving a peak accuracy of 83.33% at K-values of 3, 5, and 7. This outcome confirms the capability of GLCM-extracted texture features to distinguish between the two hijab types and highlights the potential application of this system in the field of Muslim fashion.