Camera based hand recognition for the graphically extended hand

In this paper, we propose a manipulation method for 10 foot-UI system called `Extended Hand' which is a graphical extension of a user's hand. In Extended Hand system, a combination of the forces applied to each finger control the position and the posture of Extended Hand. While the system detects both the pressing of each finger and its direction in video processing, it recognizes the translational direction and identify gripping motion and translation by using each finger displacement. In experiments, the system can recognize the user's translational direction at the directional error within 1.2 deg. and the dispersion within 7.5 degree, also identify translation and gripping with 96% identification rate.