Fast color fiducial detection and dynamic workspace extension in video see-through self-tra.pdf

Fast Color Fiducial Detection and Dynamic Workspace Extension in Video See-through Self-Tracking Augmented Reality Youngkwan Cho Jun Park Ulrich Neumann Computer Science Department Integrated Media Systems Center University of Southern California Los Angeles, CA 90089, USA { ykcho | peterpan | uneumann }@ /~ykcho/star.html Abstract The registration problem is one of the major issues in Augmented Reality (AR). Fiducial tracking is gaining interest as a solution to this problem in video see-through AR because of the availability of digitized real scenes. There are several AR systems using fiducial tracking, but most of them operate in small desktop workspaces. It is difficult to apply them directly to large scale applications. The wide range of work distance and non-uniform lighting conditions make fiducial detection very difficult. Adding new fiducials requires off-line process for measuring positions of new fiducials. In this paper, we propose a fast and robust fiducial detection procedure with carefully designed color fiducials and noise analysis of digitized images. We also present a dynamic workspace extension method with on-line position determination of unknown features. Our efforts is a framework for applying AR to large scale applications. 1. Introduction In Augmented Reality (AR), virtual objects have their own volumes and locations in the user’s real world. To add virtual objects on the user’s view, AR uses two types of cameras: a real camera and a virtual camera. A real camera generates an image stream of the user’s real world and a virtual camera adds 3D graphics of virtual objects on the live image stream. Misalignment of these cameras makes virtual objects appear incorrectly on the composite images. Registration between virtual objects and the real scene is one of the major issues in AR [Azuma94] [Bajura95][Tuceryan95]. To solve the registration problem, both of the cameras should maintain the same camera pose (position and orientation) as well as int