Efficient Line and Patch Feature Characterization and Management for Real-time Camera Tracking

Abstract

One of the key problems of augmented reality is the tracking of the camera position andviewing direction in real-time. Current vision-based systems mostly rely on the detectionand tracking of fiducial markers. Some markerless approaches exist, which are based on3D line models or calibrated reference images. These methods require a high manualpreprocessing work step, which is not applicable for the efficient development and designof industrial AR applications.The problem of the preprocessing overload is addressed by the development of vision-basedtracking algorithms, which require a minimal workload of the preparation of referencedata.A novel method for the automatic view-dependent generation of line models in real-timeis presented. The tracking system only needs a polygonal model of a reference object,which is often available from the industrial construction process. Analysis-by-synthesistechniques are used with the support of graphics hardware to create a connection betweenvirtual model and real model.Point-based methods which rely on optical flow-based template tracking are developedfor the camera pose estimation in partially known scenarios. With the support of robustreconstruction algorithms a real-time tracking system for augmented reality applicationsis developed, which is able to run with only very limited previous knowledge about thescene. The robustness and real-time capability is improved with a statistical approach fora feature management system which is based on machine learning techniques.III