EKSTRAKSI FITUR PADA PENGENALAN KARAKTER AKSARA JAWA BERBASIS HISTOGRAM OF ORIENTED GRADIENT

Abstract

ABSTRAK Buku-buku kuno Bahasa Jawa memiliki konten kekayaan intelektual Indonesia seperti agama, linguistik, filosofi, mitos, pelajaran moral, hukum dan norma adat, kerajaan, cerita rakyat, sejarah, dan lain sebagainya. ABSTRACT Ancient Javanese books have Indonesian intellectual property content such as religion, linguistics, philosophy, myths, moral lessons, customary laws and norms, kingdoms, folklore, history, and so on. Not many people studied the work because it was written in Javanese script and not many understood it. To help translation of Javanese documents, translation system automation is carried out. The translation stage consists of segmentation to get characters from the written image in the Javanese script. Then each character is recognized as alphabet. And the last is to combine alphabets that have been recognized as meaningful words. Research about recognition of Javanese scripts has been carried out, such as the focus on character seg-mentation and the introduction of Javanese script. In the previous study, improvements were made to the segmentation method but still obtained the same results in terms of accuracy. In this study we propose Histogram of Oriented Gradient (HOG) method for extracting feature. The HOG method is widely used in object recognition detection on image data. The HOG method has also been used in handwriting recognition of English and Bengali characters with optimal result. In this study the accuracy of the character recognition of Javanese script is 93.3%.