Zoning Feature Extraction for Handwritten Javanese Character Recognition

Abstract

A Javanese character used to replace Javanese, Madurese, Sundanese, Balinese and Sasak texts. Later then during Dutch colonialism era, Javanese character was replaced by Latin letters. Various efforts have been made by the government to preserve cultural heritage. For example, Javanese character reading and writing skills are also taught in schools, from elementary school to junior high school, especially in Central Java and East Java. Even so, currently not many people can use Javanese character. Javanese script does have a variety of forms so it requires understanding and memorization. To make it easier to understand each character, the technology of Javanese translation into Latin letters is one of the solutions. This technology called Character Recognition. There is an important stage in character recognition called feature extraction. This stage will determine the result of the classification process. In this paper, the adoption of feature extraction methods introduced by Elima Hussain called Zoning Feature Extraction is applied. This method works by dividing an image into several zones and counts the number of black pixel from each zone. Based on the experiment, this method with the number of zone 16 can extract features of Javanese Character.