Re: HELP in Delaunay Triangulation

"John D'Errico" <woodchips@xxxxxxxxxxxxxxxx> wrote in
message <fv9tta$5r6$1@xxxxxxxxxxxxxxxxxx>...
"Sherman Tse" <man_superman_batman_007@xxxxxxxxx> wrote
in message
<fv9qtm$f0o$1@xxxxxxxxxxxxxxxxxx>...
Dear John

Sorry for the insufficient details!

Let me explain it a little bit more!

My project is to use 3D point cloud data to reconstruct
a
real 3D model back!! My project is to use the Digiclops
to
get the point cloud data of a human face.

The point cloud data has X Y Z coordinates, RGB value
of
each vertex.

Would the Delaunay Triangulation be appropriate
appraoch!??

I have used the "surfi" in FEX . the Triangulation is
not
what I want, I want to create another Triangualtion
method.
I want to create triangles from every point, so that I
can
map the color of each vertex and then interpolate the
color
to reconstruct a real model back.

I have called the "delaunay" and "delaunay3" in the
matlab.
however, the triangles are too big that the model is
accurate enough to show the contour of the human face.

What should I do, Could you please help me to create an
algorithm??PLease give me some inspiration!!

The problem is not that the triangles are
too big. The problem is that your point
"cloud" does not in fact arise from a convex
object. A face would have indentations that
are covered over by those large facets,
since a delaunay tessellation will always
be convex. Delaunay triangulations are
simply not the proper tool or these
problems.

John


Then what should I do?
.