Hier finden Sie wissenschaftliche Publikationen aus den Fraunhofer-Instituten.

PTrack: Introducing a novel iterative geometric pose estimation for a marker-based single camera tracking system

: Santos, P.; Stork, A.; Buaes, A.; Jorge, J.


Fröhlich, B. ; Institute of Electrical and Electronics Engineers -IEEE-:
IEEE Virtual Reality 2006. Proceedings : Alexandria, Virginia, USA, March 25-29, 2006
Piscataway: IEEE Computer Society, 2006
ISBN: 1-4244-0223-9
ISBN: 1-4244-0224-7
Virtual Reality Conference (VR) <13, 2006, Alexandria/Va.>
Fraunhofer IGD ()
Augmented reality (AR); Virtual reality (VR); camera tracking; optical tracking system

Inside-out tracking for augmented reality applications continues to present a challenge in terms of performance, robustness and accuracy. In particular mobile augmented reality applications are in need of tracking systems which can be wearable and do not cause a high processing load.
In this paper we introduce a novel iterative geometric method for pose estimation from four co-planar points and we present the current status of PTrack, a marker-based single camera tracking system benefiting from this approach. The system uses an IDS uEye [1] camera, equipped with infrared flash strobes and infrared pass filter, which acquires grayscale images and sends them to a computer where an image pre-processing algorithm identifies potential projections of retro-reflective markers. Our novel pose estimation algorithm identifies possible labels composed of markers in a 2D post-processing using a divide-and-conquer strategy to segment the camera's image space and attempts an iterative geometric 3D reconstruction of position and orientation in camera space. In the end successfully reconstructed labels are compared to a database for identification. Their 6 DoF data as well as their ID is made available to applications through OpenTracker [2] framework. To assess the performance of our approach we compared PTrack to ARToolKit [3] - which has been adapted to work with the same camera - and final results show that pose estimation is more accurate and precise, in both translation and rotation.