Self-Localization in Scanned 3DTV Sets

Future 3D Television applications will offer the viewer to freely choose his viewpoint during transmission. A lot of research in the field of 3DTV therefore concentrated on capturing photo-realistic 3D models of studio or movie sets. In this paper, however, we concentrate on the problem of self-localization within such scenes. As input we expect a 3D model of an arbitrary environment. Therein, we are able to localize a low-cost portable sensor-system based on a 3D time-of-flight camera. Point clouds acquired with this system from arbitrary viewpoints are registered to the given model in order to estimate its position and orientation in the scene.