Real-time Gesture Recognition for the Control of Avatar

There are many attempts to use the gestures for the natural human computer interface. Most of systems use the tracking devices to capture the human motion and display. In this paper, we propose a gesture recognition system for natural interface between the participant and his own avatar in virtual environment only using two polhemus sensors attached on the arm. Our system has two modes for the control of avatar. One is tracking gestures. In this mode, tracking data is used only to simulating the motion of the participant’s one arm. The other is the interpretation of gestures. In the interpretation mode, the participant can make his avatar walk or run by the speed corresponding to the velocity of gesture and command his avatar to work properly by natural gesture. This gesture interpretation is done in real-time and may give an immediate response in virtual environments. Therefore, our system provides more natural interface in virtual environments by the gesture.