In this paper, we present a new algorithm for reconstructing an environment from images recorded by multiple calibrated cameras. Multiple camera systems challenge traditional stereo algorithms in many issues including view registration, selection of commonly visible image parts for matching, and the fact that surfaces are imaged differently from different viewpoints and poses. On the other hand, multiple cameras have the advantage of revealing surfaces at occluding contours and covering wide areas. The presented algorithm makes no assumption on camera loci and outputs an occupancy voxel grid, with occupied voxels being accompanied by a surface normal. It is correlation-based, however, outperforms the conventional correlation-based approach in reconstruction quality. It is highly parallelizable, and most importantly, is robust against artifacts due to camera registration errors that are typically encountered when using multiple cameras.