Transputer-based manipulation of computed-tomography images

The Millar polyhedron is introduced as a technique which makes use of volume coherence in the image data and thus, in general, reduces the quantity of data which must be stored and manipulated. It is demonstrated how this technique aids the construction of an octree by generating data in a way which directly transfers to the terminal nodes of the octree. Image production also utilises the inverse Millar polyhedron. This modelling methodology is then applied to the parallel slices produced by a CT-scanner. Timed results are given for the production of slices from a three-dimensional model built using an implicit octree data structure. These demonstrate how the transputer approach exploits the inherent concurrency in the production of such images