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

Abstract

In this paper we present an improved color-based planar¯ducial marker system. Our framework provides precise and robust full 3D pose estimation of markers with superior accuracy when compared with many¯ducial systems in the literature, while color information encoding enables using over 65 000 distinct markers. Unlike most color-based¯ducial frameworks, which requires prior classi¯cation training and color calibration, ours can perform reliably under illumination changes, requiring but a rough white balance adjustment. Our methodology provides good detection performance even under poor illumination conditions which typically compromise other marker identi¯cation techniques, thus avoiding the evaluation of otherwise falsely iden-ti¯ed markers. Several experiments are presented and carefully analyzed, in order to validate our system and demonstrate the signi¯cant improvement in estimation accuracy of both position and orientation over traditional techniques.

Optimality of black to white ratio and information complexity for robust marker recognition

“…They mentioned that their markers may be misdetected due to low quality cameras. The paper also concluded that larger quantity of distinct markers can be achieved by increasing their size [16]. However there is no experimental proof given.…”

“…The square shape is better for easy pose calculation [14]. Black and white markers are robust in various light intensity levels and are easy to print [15] whereas color markers allows a comparatively large number of distinct markers [16]. Markers of ARTag are also square in shape where the marker is divided into 10x10 segments [17].…”

“…Apart from AR applications, fiducial markers are used in other areas including robot localization, medical, television, industrial, video conferences, and human machine interactions [16], [18]. Markers are used in indoor AR, hand-held objects for showing user pose to the camera, message tags to obtain some feature from marker, or general pose calculation for industrial setting [10].…”

“…The requirements for high quality markers are that: the marker should be distinct from the surrounding, unique in existing library of markers, passive (not coated with electronic substances), quickly detectable and effective in low light and noisy environment using a robust image processing algorithm [10]. The use of black and white markers enable easy detection in various lighting conditions [14], whereas using more colors support a greater number of unique markers [16]. The main challenge in AR applications is to achieve high speed and accuracy with low cost and minimal changes in the applications [3].…”

“…Similarly in [17], new type of markers (called diagonal connected component markers) are introduced and the results are compared with ARToolKit markers in terms of detection time. V. F. da Camara et al [16] use color based markers to create up to 65000 unique markers with comparatively greater accuracy. They mentioned that their markers may be misdetected due to low quality cameras.…”