Automatic batik motifs classification using various combinations of SIFT features moments and k-Nearest Neighbor

Abstract

Batik cloth is Indonesia's national heritage. Across the archipelago, there are numerous patterns and motifs of batik, each having its own meaning and cultural significance. In this paper, we present the results of our investigation of various combinations of SIFT features moments used in automatic classification of batik motifs. The classification method used in this paper is the k-Nearest Neighbor. Our experiments show that the best performance of the system is obtained using feature vectors of length 7, yielding a classification accuracy rate of 31.43% for 7 classes of batik motifs with no batik motif classes having zero classification accuracy rate. Furthermore, our experiments suggest that the feature moment that seems to be the best for the classification process is the μc, while the feature moment that seems to hinder the classification process is the σc2.