Feature Extraction from Smartphone Images by Using Elliptical Fourier Descriptor, Centroid and Area for Recognizing Indonesian Sign Language SIBI (Sistem Isyarat Bahasa Indonesia)

Abstract

Sistem Isyarat Bahasa Indonesia (SIBI) is the official sign language in Indonesia. This research aims to create a translator for SIBI to be installed on a smartphone. The translator will analyze gestures for inflectional words, which are root words combined with prefixes, and/or suffixes. The feature extraction method that was used in this research is Elliptical Fourier Descriptor (EFD), additionally centroid and area data were added to retain information on hand orientation, position and shape. The extracted features will be fed into, a Long Short-Term Memory (LSTM) model which will then recognize gestures into text. The method used in this research produced 99% accuracy for root word gestures, 71% accuracy for prefix gestures, 86% accuracy for suffix gestures.