Posing to the camera: Automatic viewpoint selection for human actions

Abstract

In many scenarios a scene is filmed by multiple video cameras located at different viewing positions. The difficulty in watching multiple views simultaneously raises an immediate question - which cameras capture better views of the dynamic scene? When one can only display a single view (e.g. in TV broadcasts) a human producer manually selects the best view. In this paper we propose a method for evaluating the quality of a view, captured by a single camera. This can be used to automate viewpoint selection. We regard human actions as three-dimensional shapes induced by their silhouettes in the space-time volume. The quality of a view is evaluated by incorporating three measures that capture the visibility of the action provided by these space-time shapes. We evaluate the proposed approach both qualitatively and quantitatively. © 2011 Springer-Verlag Berlin Heidelberg.