Klasifikasi Citra Keris Bali Menggunakan Ektrasi Fitur Gray Level Co-Occurrence Matrix Dengan Klasifier KNN (K-Nearest Neighbor)

Abstrak

Pamor is a distinctive pattern on the blade of a keris formed through folded metal forging techniques, carrying aesthetic and philosophical values in Balinese culture. Manual identification of pamor often leads to errors due to the similarity of patterns and names among its types. This study aims to develop an automatic classification system for Balinese keris pamor using digital image processing. The dataset was obtained from the Bali Keris Museum consisting of 26 original images, which were pre-processed through cropping, grayscale conversion, and resizing. To balance the dataset, image augmentation was applied by rotating 45°, 90°, and 180°, resulting in 250 images evenly distributed across five pamor classes: tiban, miring, puntiran, tambal, and gedheg. Texture features were extracted using the Gray Level Co-Occurrence Matrix (GLCM) with parameters of contrast, correlation, energy, and homogeneity. These features were then classified using the K-Nearest Neighbor (KNN) algorithm with K values of 7, 9, and 11. The results indicate that K = 7 achieved the highest accuracy of 80%, while K = 9 obtained 78% and K = 11 decreased to 68%. This finding confirms that the selection of the K parameter significantly influences classification performance, and that the combination of GLCM and KNN can effectively support the automatic and objective identification of Balinese keris pamor

Identifier

• OpenAlex: https://openalex.org/W7131076190
• DOI: https://doi.org/10.31598/sintechjournal.v8i2.1925