Recognition of sign language with an inertial sensor-based data glove

• Published in: Technology and health care Vol. 24 Suppl 1; no. s1; pp. S223 - S230
• Main Authors: Kim, Kyung-Won, Lee, Mi-So, Soon, Bo-Ram, Ryu, Mun-Ho, Kim, Je-Nam
• Format: Journal Article
• Language: English
• Published: Netherlands 08.12.2015
• Subjects: Algorithms; Hand; Humans; Movement; Pattern Recognition, Automated - methods; Republic of Korea; Sign Language; Translating; Republic of Korea; inertial sensor; Data glove; sign language recognition; accelerometer
• DOI: 10.3233/THC-151078
• ISSN: 0928-7329

Communication between people with normal hearing and hearing impairment is difficult. Recently, a variety of studies on sign language recognition have presented benefits from the development of information technology. This study presents a sign language recognition system using a data glove composed of 3-axis accelerometers, magnetometers, and gyroscopes. Each data obtained by the data glove is transmitted to a host application (implemented in a Window program on a PC). Next, the data is converted into angle data, and the angle information is displayed on the host application and verified by outputting three-dimensional models to the display. An experiment was performed with five subjects, three females and two males, and a performance set comprising numbers from one to nine was repeated five times. The system achieves a 99.26% movement detection rate, and approximately 98% recognition rate for each finger's state. The proposed system is expected to be a more portable and useful system when this algorithm is applied to smartphone applications for use in some situations such as in emergencies.