Feature Extraction for Indonesian Sign Language (SIBI) Using Leap Motion Controller

Abstract

Communication is an essential needs for people. Generally, people communicate verbally. However, some people have a deficiency of not being able to communicate verbally like deaf and mute people. They use sign language to communicate. Many research has been done to translate or recognize sign language. Key to sign language translation is hand gesture recognition. In this paper, Indonesian Sign language (SIBI) Recognition System is proposed. This system use Leap Motion Controller device and new feature extraction method to obtain more accurate gesture data. Input data taken from Leap Motion is position data within the time range t This position data is calculated by relative coordinate so as to get the vector feature for a movement. Data classification processed by k-Nearest Neighbor (k-NN) and Support Vector Machine (SVM). This proposed system using new feature extraction method gives an average accuracy level of sign language recognition equal to 95.15% by using k-NN classification and 93.85% by using SVM classification.