Natural Feature Tracking for Extendible Robust Augmented Realities

Vision-based tracking systems for augmented reality often require that artificial fiducials be placed in the scene. In this paper we utilize our approach for robust detection and tracking of natural features such as textures or corners. The tracked natural features are automatically calibrated to the fiducials that are used to initialize and facilitate normal tracking. Once calibrated, the natural features are used to extend the system’s tracking range and to stabilize the tracked pose against occlusions and noise. The emphasis of this paper is the integration of natural feature tracking with fiducial tracking to increase the range and robustness of vision-based augmented reality tracking.