Analysis of Color and Texture Features for Samarinda Sarong Classification

Abstract

Samarinda sarong or Tajong Samarinda is a traditional woven fabric originating from Samarinda, East Borneo, Indonesia. It is made through a weaving process using a loom called a Gedokan (a traditional machine). Unfortunately, many Samarinda people still lack knowledge regarding the type of Samarinda sarong; hence they cannot recognize it. Therefore, an automatic method of image processing-based needed to recognizing and classifying the motif of Samarinda sarong. This method requires appropriate and discriminatory features to obtain the optimal classification results. This work aims to analyze color and texture features to produce discriminative features. The color features used are color moments applied on RGB and HSV color spaces, while texture features were extracted using Gray Level Co-occurrence Matrix (GLCM). Subsequently, those features were reduced using correlation-based feature selection (CFS) followed by applying the Support Vector Machine (SVM) classifier. The dataset used consists of 150 sarong images (50 Belang Hata, 50 Belang Negara, and 50 Kuningsau). The method performance successfully achieved the accuracy of 100% using only 10 color features from a total of 34 features.