A novel scene reconstruction technique is presented, different from previou
s approaches in its ability to cope with large changes in visibility and it
s modeling of intrinsic scene color and texture information. The method avo
ids image correspondence problems by working in a discretized scene space w
hose voxels are traversed in a fixed visibility ordering. This strategy tak
es full account of occlusions and allows the input cameras to be far apart
and widely distributed about the environment. The algorithm identifies a sp
ecial set of invariant voxels which together form a spatial and photometric
reconstruction of the scene, fully consistent with the input images. The a
pproach is evaluated with images from both inward-facing and outward-facing
cameras.