If you're talking about automatic recreation of 3d objects from a stereo-pair, that would be the billion dollar prize that every government would love to have. However, if you have planar images (x-y) for which you have some idea of alignment in volumetric space (z axis), you could import intensity values into a voxel cube space, or use some of the OpenGL rendering techniques. If you import into the volume cube, understand that there may be major memory requirements. To fill in voids, you can potentially use the convol function, or create your own technique. If you happen to work with any astronomers, they have a couple of unique reconstruction techniques that involve spectroscopic imagery (for position-velocity) that works with standard astrophotography for 3d volumetric reconstruction.
Hope this helps.
-- Matt Altman
|