Integration of Vision and Inertial Sensors for Home-based Rehabilitation

This paper introduces a real-time tracking system of human arm motion, specifically intent to be used for homebased rehabilitation. Both vision and inertial sensors are employed in this system to track the arm movement in 3D and in real time. Different data modalities from two sensors are fused by using arm structure relationship and geometry information. The occlusion problem, which is one of the most difficult issues in most of vision only tracking systems, is well dealt with by combining the geometry information of the subject and multiple sensors in our proposed system. The experimental results show our system can track the 3D arm motion in real time and has acceptable accuracy, compared with the commercial markerbased system, CODA.