A Combination of Statistical Extraction and Texture Features Based on KNN for Batik Classification

Batik has various motives in every region in Indonesia. Various types of batik motifs have characteristic patterns that differ from one another. There are many computer systems that can detect various types of batik motifs to classify the types of batik motifs. One of the systems that have been used to solve the problem is K Nearest Neighbor (KNN) combined with Gray Level Co-Occurrence Matrix (GLCM). The dataset used is an image of a batik motif, amounting to 300 images divided into 50 classes. These images are RGB (Red Green Blue) color with the JPG format. Confusion Matrix was carried out to do the classification accuracy testing. There are five stages of the research process: the image input process, pre-processing, feature extraction, classification, and testing. The highest accuracy results are 96% at the value of k = 1, while the lowest accuracy results are 86% at the value of k = 6. The accuracy results obtained from this study are much higher than the previous research.