Hand-Tracking Framework using Three-dimensional Tracking Ellipsoid

Abstract

In this paper, we present a hand-tracking system. Our system uses a depth-capture device to acquire the 3D geometry of a user's bare hand, such that the user's hand position and gesture information can be estimated from the depth cues. In particular, we build an ellipsoidal tracking object and restrict the hand tracking to within the ellipsoidal area to avoid the diverse problems occurring in conventional object detection/tracking systems. This tracking area is actively controlled according to the statistics of a user's hand gesture and motion history. The proposed system aims to be used in diverse systems that operate based on the natural user interface framework. The experimental results verify the accuracy and efficiency of the proposed system.