CGAL users discussion list

[Thijs van Lankveld <>]

On 24/06/2014 18:02, colonel_panic wrote:
> Thank you both for the replies.  My algorithm is similar to that which
> produced these results:
> http://www.mage.fst.uha.fr/maillot/recherche/images/LDA.gif
>
> I need to generate a delaunay triangulation from a point set to start with
> and then refine it - the outcome could be a more general simplicial comple.
> I would like to be able to remove facets though - an example of where this
> would be useful would be reconstructing a shape from a point set which has
> several holes or hollows.  the 3d triangulation generates many cells which
> go across the hollow and I would like to remove the cells which are in empty
> space while keeping the ones which are on the surface of the shape.
>
> I have thought about testing each facet and then printing each one which
> meets the criteria to be included but this seems like a sluggish solution.
> I was asking here first to see if there was an easier/more elegant way to do
> it :)
If I understand your application correctly, you never need edges/facets 
that were not in the original triangulation.
In this case, it is probably best to mark the cells/facets/edges that 
you want to display. The connectivity of the cells should then not 
change, just the marks while you iterate. In other words, you don't 
remove any cells/facets/edges, you just choose whether to ignore them.

How you store these marks is up to you.
See 
http://doc.cgal.org/latest/Triangulation_3/classCGAL_1_1Triangulation__cell__base__with__info__3.html 
for a cell type that could store these marks if you can encode them in 
one value.
You could also implement your own cell_base subclass, but this does 
require significantly more effort. See also: 
http://doc.cgal.org/latest/Triangulation_3/index.html#title7

> --
> View this message in context: 
> http://cgal-discuss.949826.n4.nabble.com/Removing-edges-and-facets-from-a-3d-triangulation-tp4659472p4659475.html
> Sent from the cgal-discuss mailing list archive at Nabble.com.