On Jul 14, 3:29=A0am, Hans-Bernhard Br=F6ker <HBBroe...@[EMAIL PROTECTED]
>
wrote:
> Keith.Lv wrote:
> > The input triangles are the surface of a bounded object, and they are
> > very similar to a manifold but some edges are shared by more than 2-
> > triangles.
>
> That statement is somewhat self-contradictory. =A0Triangles that "are"
th=
e
> surface of a bounded volume, would automatically form a manifold. =A0So
> either they not really "are" the surface (but more-or-less just lie on
> it) or this recipe
>
> > My previous idea is to delete some illegal triangles, and generate a
> > manifold mesh, and then fill the small holes on the surface.
>
> must be flawed, because the real defects are not extra triangles, but
> rather that edges/vertices have been classified as "the same" a bit too
> eagerly. =A0I.e. you should probably split edges (and vertices) instead
o=
f
> killing entire triangles.
I'm sorry for that I havn't describe the problem clearly and make you
confused.
The object described by the mesh is bounded, but doesn't mean the mesh
is. The mesh is get from a reconstruction algorithm. Since the real
object is bounded, I thought the mesh should be manifold. But in fact,
the reconstruction algorithm will generate a non-monifold in sharp
edges, the triangles near the sharp edges will connected like the
letter "A", and make the mesh non-manifold near this area. My task is
to delete some triangles to change "A" to "/\", or just let this part
just be a hole. The output needn't be perfect, bust must be manifold.
The next step in our application will fix the left problems.
I've write a algorithm by insert the triangles as many as possible in
the order of breath first search. It's not a perfect algorithm, but it
works and meets our requirement.
Thank you for your reply.
|
