On the Design and Evaluation of a Precise Scalable Fiducial Marker Framework

In this paper we present a planar fiducial marker system to be used with color cameras. Our system provides precise and robust full 3D pose estimation of the markers with superior accuracy when compared with many fiducial systems, while color information is used to provide more than \n{65000} distinct markers. In contrast with most color-based fiducial frameworks, ours requires no prior classification training nor color calibration other than a rough white balance adjustment and can perform reliably under illumination changes. Finally, we also provide means of detecting when poor illumination conditions will compromise marker identification, thus avoiding to evaluate a false marker identification. We present several experiments that show significant improvement in accuracy of estimation of both position and orientation when compared with traditional techniques.