Roundness and Eccentricity Feature Extraction for Javanese Handwritten Character Recognition based on K-Nearest Neighbor

Abstract

Javanese script ha-na-ca-ra-ka is a relic of the ancestors of the nation Indonesia. Javanese script has 20 basic character types, each character has a complexity of writing is quite complicated because it is very different from the alphabet letter. So that each character is difficult to recognize and learn. Recognition algorithm can be applied to the computer to recognize Javanese script character. K-Nearest Neighbor (KNN) is a classification algorithm that can be used for character recognition. The recognition process needs the help of feature extraction. This research proposes the extraction of roundness and eccentricity features to characterize the character of Javanese characters. To get a significant result, the training data and test data from Javanese script image written in hand done some preprocessing phase. Some of the steps are cropping the image, converting to a negative image, median filtering, binary, and dilation. The amount of data used is 240 handwritten Javanese characters, consisting of 40 test data and 200 training data. Experimental results of the proposed method obtained an accuracy of 87.50%.