Development and experimental evaluation of an automatic marker registration system for tracking of augmented reality

In order to apply augmented reality in plant maintenance activities it is necessary to use real-time high accuracy tracking technology. One of the most efficient tracking methods is using paper-based markers and computing the relative position and orientation between a vision sensor (camera) and the markers through image processing and geometry calculations. In this method, the 3D-position of each marker is needed before tracking, but it is inefficient to measure all the markers manually. In this study, an automatic marker registration system was developed so as to measure the 3D-position of each marker automatically. The system is composed of a camera, a laser rangefinder and a motion base, which is used to control the pose of the laser rangefinder. A computer, connected to them, is used for controlling the system and for data transport. The results of the experimental evaluations show that the measurement takes about 21 seconds per marker and that the Root Mean Square Error (RMSE) of the position measurements is 3.5 mm. The feasibility evaluation of the system was conducted in Fugen nuclear plant. The results show that the system can largely reduce the preparatory workload of an AR application in a