Pengenalan Aksara Jawi Tulisan Tangan Menggunakan Freemen Chain Code (FCC), Support Vector Machine (SVM) dan Aturan Pengambilan Keputusan

• Main Authors: Safrizal, Fitri Arnia, Rusdha Muharar
• Format: Article
• Language: English
• Published: Universitas Andalas 2016-03-01
• Series: Jurnal Nasional Teknik Elektro
• Subjects: Jawi script; handwriting; FCC; SVM; decision rules
• Online Access: http://jnte.ft.unand.ac.id/index.php/jnte/article/view/185
• ISSN: 2302-2949; 2407-7267

Jawi is one variant of Arabic script consists of 35 characters. Some of Jawi characters have the same main shape, but different number of dots in different location. Thus, recognition process of Jawi characters can be done by performing a classification based on the main shape. In recognition process, feature extraction plays an important role. In this research, Freeman Chain Code (FCC) was used as feature extraction and Support Vector Machine (SVM) as classifier. Then we apply the decision rules to classify SVM result into Jawi characters. FCC is used to represent the boundary of Jawi characters into a chain code. Then the chain code is used by SVM to classify the characters into 19 groups. Feature of location and the number of dots are used by decision rules to classify the groups into Jawi characters. The Jawi characters are handwritten and generated by 10 writers from different backgrounds and ages. The recognition rate of this research was 80.00%.