3D-TV LDV content generation with a hybrid ToF-multicamera RIG

In this paper we introduce an approach for automatic generation of Layered Depth Video (LDV) content for 3D television (3DTV) purposes. The proposed approach relies on a capturing system consisting of two time of flight cameras (ToF) and five color cameras. The LDV content is generated for the central color camera, based on the depth images provided by the ToF cameras and texture information provided by the color cameras. Through different viewpoints of the ToF cameras and color cameras, we are able to reconstruct depth and texture information not only for foreground objects, but also in the occluded areas. The estimated information is then combined to one consistent LDV representation.