CGAL users discussion list

[Cristobal Navarro <>]

On Mon, May 21, 2012 at 2:47 PM, Daniel Duque [via cgal-discuss]

<[hidden email] </user/SendEmail.jtp?type=node&node=4649966&i=0>> wrote:

   El 21/05/2012 19:49, neoideo escribió:
    > Hello,
    >
    > Suppose i have a 2D triangulation loaded into cgal, from a file
   mymesh.off
    > and the triangulation
    > is not Delaunay.
    >
    > Then, my question:
    > Is there any method (already implemented in CGAL) that improves the
    > triangulation based on the Delaunay criteria??

   Perhaps you can just read the positions of each vertex, not their
   connectivities, then use them
   to build a new Delaunay triangulation.


 Yes, indeed your solution works. However, i needed a faster method
that could reuse the connectivity
so i can compare performance and correctness against another method not
in cgal.
At least, with your solution,
i can compare how exact is the other method against cgal delaunay mesh.
But performance should be very slow on cgal side.

Hi

Unless your input triangulation is almost Delaunay, there is a fair chance that recomputing the whole Delaunay triangulation is faster than transforming your triangulation so that it becomes Delaunay.
Computing a 2D Delaunay triangulation from scratch with CGAL is roughly 1 second for 1 million points.

Best,
--
Monique Teillaud
INRIA Sophia Antipolis - Méditerranée
http://www.inria.fr/sophia/members/Monique.Teillaud/

Your information is very useful
today i will have some results where i can check this fenomenom
thank you