A six-DOF motion tracking system for markered environment

In this article, we describe a system that estimates the position and orientation of a calibrated camera. For better estimation accuracy, artificial markers made up of infrared LEDs are placed in certain pattern onto the ceiling of a room. A 3D reconstruction process is first performed offline so that the coordinates of all LEDs within a unified world frame are determined. This process, based on incremental optimization, has proved to give more accurate result than the traditional global iterative optimization. During real-time tracking, the infrared lens filter of the camera helps to filter out to a large degree of the visible light, thus the system is fairly robust and does not suffer the impact of the illumination changes. To reduce the accumulated error, an inertial sensor is fused with the camera based estimation and its signal acts as the control vector in the Extended Kalman Filter. The experimental result demonstrates the speed and robustness of our proposed system.