Enhancing voxel carving by capture volume calculations

Many multi camera based approaches on dense volumetric 3d reconstruction depend on the correctly chosen often manually defined reconstruction volume and a reliable foreground/background segmentation process. We present an algorithm to automatically calculate the 3d polyhedron of the reconstruction volume seen by all cameras of a calibrated multi camera setup by utilizing the calibration data. The information from this polyhedron can not only be used to define the reconstruction volume but also to improve the reconstruction process significantly in terms of runtime and/or quality of reconstruction as presented in the evaluation section, in which the algorithm has been successfully tested on real life data.