Deteksi Sarung Samarinda Menggunakan Metode Naive Bayes Berbasis Pengolahan Citra

Abstract

Samarinda sarong is one of the cultural treasures in the form of cloth from Samarinda, East Kalimantan. It has a characteristic in the form of a square motif with a unique color combination. However, several people do not know the difference between a Samarinda sarong and a non-Samarinda sarong because the Samarinda sarongs may have a similar motif or color to a non-Samarinda sarong. This study aims to develop a Samarinda sarong detection method to distinguish between the sarong of Samarinda and non-Samarinda. The detection of the Samarinda sarong was carried out based on two features: color and texture. The feature extraction of color was applied using color moments and Gray Level Co-Occurrence Matrix (GLCM) for texture. The classification was implemented using the Naive Bayes method. The dataset used consists of 250 sarong images (150 Samarinda sarong images and 100 Non-Samarinda sarong images) divided into training and test data. It was divided using percentage split and cross-validation. The test results show the implementation of the color moments, GLCM, and Naive Bayes methods using a percentage split (70%) produce the best accuracy of 0.987 compared to using cross-validation (K=10) with an accuracy of 0.984. The difference may occur because the number of training and testing data used on percentage split and cross-validation is different. Moreover, the sarong images used on training and test data were chosen randomly.

Keywords

• samarinda sarong, color moments, glcm, feature extraction, naive bayes, classification