STag: A Stable Fiducial Marker System

In this paper, we propose STag, a fiducial marker system that provides stable pose estimation. The outer square border of the marker is used for detection and pose estimation. This is followed by a novel pose refinement step using the inner circular border. The refined pose is more stable and robust across viewing conditions compared to the state of the art. In addition, the lexicographic generation algorithm is adapted for fiducial markers, and libraries with various sizes are created. This makes the system suitable for applications that require many unique markers, or few unique markers with high occlusion resistance. The edge segment-based detection algorithm is of low complexity, and returns few false candidates. These features are demonstrated with experiments on real images, including comparisons with the state of the art fiducial marker systems.