Multi-markers 3D tracking algorithm in a video motion capture system

In a video motion capture system based on multi-markers, it is difficult to match and track these markers because their image features are quiet similar. This paper proposes a novel 3D tracking algorithm to track markers with similar image features. Several extended Kalman filters are used to predict each marker’s 2D image coordination and 3D position. The 3D stereo matching method and epipolar geometry restriction are combined to eliminate the incorrect matching points and solve the occlusion in the procedure of stereo tracking. Experimental results demonstrate that the proposed algorithm can track multiple markers with similar image features exactly in binocular stereo vision.