<![CDATA[The limited human resources who are able to read Sasak script palm leaves are the main motivation for developing a transliteration tool for Sasak script images on palm leaves. By using the Reduced Support Vector Machine or RSVM algorithm as one of the classification methods, transliteration efforts can be facilitated with maximum results. The principle of the RSVM method in classifying objects by separating two different classes using a hyperplane has been proven to be able to produce maximum accuracy performance in this study. The research data in the form of Sasak script images resulting from the palm leaf image segmentation process that has been divided into 18 classes. The feature extraction algorithm used is Intensity of Character (IoC) with window sizes of 3x3, 4x4, and 5x5 and 3-Fold, 5-Fold, 7-Fold data Imbalance. The test results at the RSVM classification stage using the Linear Kernel, Polynomial Kernel, Radial Basic Function (RBF) and One against One modeling on the 18 classes tested, where each class contains 20 handwritten Jejawen Sasak script image data on palm leaves, were recorded to produce the highest accuracy, which was 93.6%.]]>
Copyrights © 2025
