Fiducial Marker Based on Projective Invariant for Augmented Reality

Fiducial marker based Augmented Reality has manyapplications. So far the inner pattern of the fiducial marker is alwaysused to encode the markers. Thus a large portion of the fiducial markerimage is used for encoding instead of providing corresponding featurepoints for pose accuracy. This paper presents a novel method whichutilizes directly the projective invariant contained in the positionalrelation of the corresponding feature points to encode the marker. Theproposed method does not require the region of pattern image forencoding any more and can provide more corresponding feature points sothat higher pose accuracy can be achieved easily. Many relatedapproaches such as cumulative distribution function, reprojectionverification and robust process are proposed to overcome the problem ofsensibility of the projective invariant. Experimental results show thatthe proposed fiducial marker system is reliable and robust, and canprovide higher pose accuracy than that achieved by existing fiducialmarker systems.