Real-time Camera Tracking for Augmented Reality Applications : Efficient Line and Patch Feature Characertization and Management

ISBN-13: 9783639107395 / Angielski / Miękka / 2009 / 144 str.

Harald Wuest

Opis

One of the key problems of augmented reality is thetracking of the camera position and viewing directionin real-time. Current vision-based systems mostlyrely on the detection and tracking of fiducialmarkers. Some markerless approaches exist, which arebased on 3D line models or calibrated referenceimages. These methods require a high manualpreprocessing work step, which is not applicable forthe efficient development and design of industrial ARapplications.The problem of the preprocessing overload isaddressed by the development of vision-based trackingalgorithms, which require a minimal workload of thepreparation of reference data.A novel method for the automatic view-dependentgeneration of line models in real-time is presented.The tracking system only needs a polygonal model of areference object, which is often available from theindustrial construction process.Point-based methods which rely on optical flow-basedtemplate tracking are developed for the camera poseestimation in partially known scenarios. Therobustness and real-time capability is improved witha statistical approach for a feature managementsystem which is based on machine learning techniques.

O autorze

Harald Wuest received his diploma degree in computer engineering
from the University of Mannheim in 2004. Since 2004 he has been a
research fellow at the Fraunhofer Institute for Computer
Graphics in Darmstadt. In 2008 he received his PhD in computer
science from the University of Darmstadt, Germany.