Classification Of Batik Images Using Multilayer Perceptron With Histogram Of Oriented Gradient Feature Extraction

Abstract

Batik is one of the hereditary cultural heritages which has a high aesthetic value and a deep philosophy. Batik is one of the cultural icons from Indonesia which was awarded as a cultural heritage from UNESCO on October 2, 2009. Currently, Indonesian batik has various types of motifs and different patterns, which are scattered in Indonesia with their names and meanings. The many batik patterns in Indonesia make it difficult to identify batik motifs, especially for the common people. To overcome this problem, it is necessary to have a batik classification system with a high level of accuracy, so that everyone can recognize the batik pattern easily. In this study, the Histogram Of Oriented Gradient method was used as a feature extraction process to obtain batik density and Multilayer Perceptron as a classification method to determine the level of accuracy. The result of the level of accuracy obtained for each batik motif has a different level of accuracy. This is because the batik motifs have unique patterns and shapes that are not specific. The level of income obtained was 83.4 %. From the results of the study concluded that the use of the Histogram Of Oriented Gradient method as a feature extraction method, the Multilayer Perceptron as a classifier can be applied for image classification on batik.