Abstract | Marker patterns or beacons can be added to scenes or objects and detected automatically in camera imagery. The detection of markers in images is used for tracking position for applications such as Augmented Reality and robot navigation. Recognizing landmarks with passive vision simplifies the hardware and lowers the cost as compared to position tracking systems that use lasers, SONAR or RF based methods. A marker system comprises a set of marker patterns, usually bitonal black and white planar images, and corresponding computer vision algorithms to recognize them in images where the markers are in view. This paper details a new marker system that uses a special pseudo-random sequences encoded into linear patterns. These linear patterns can be recognized with merely a portion visible, removing the reliance on quantized boundaries as with other marker systems. The theory is explained, descriptions are given of software written applying PLIM, and two specific application shown where these linear markers are used for linear positioning of a gantry and robot navigation with panoramic optics. |
---|