Skip to Content.
Sympa Menu

cgal-discuss - Re: [cgal-discuss] 3D mesh from segmented image

Subject: CGAL users discussion list

List archive

Re: [cgal-discuss] 3D mesh from segmented image


Chronological Thread 
  • From: Phil Weir <>
  • To:
  • Subject: Re: [cgal-discuss] 3D mesh from segmented image
  • Date: Tue, 13 Oct 2015 14:14:58 +0100
  • Authentication-results: mail2-smtp-roc.national.inria.fr; spf=None ; spf=Pass ; spf=Pass
  • Ironport-phdr: 9a23:TbIVxhJ3KbQdYzovydmcpTZWNBhigK39O0sv0rFitYgUKfjxwZ3uMQTl6Ol3ixeRBMOAu64C27ed7/+ocFdDyKjCmUhKSIZLWR4BhJdetC0bK+nBN3fGKuX3ZTcxBsVIWQwt1Xi6NU9IBJS2PAWK8TWM5DIfUi/yKRBybrysXNWC04LniqvupNX6WEZhunmUWftKNhK4rAHc5IE9oLBJDeIP8CbPuWZCYO9MxGlldhq5lhf44dqsrtY4q3wD86Fpy8kVWqrze+E0TKdTES89G2Ez/szi8xfZHiWV4X5JamUXllJ3DgnfpEXiXpb19DT3t/BV3jSfNIv3QOZnCnyZ8653RUqw2288PDkj/TSPhw==

On 10/13/2015 01:54 PM, Adrien Leygue wrote:
Dear Cgal users,

I am trying to mesh a 3D segmented image using as domain oracle a « Labeled_image_mesh_domain_3 ».

The mesh is very nice, except close to the image surface where the mesh does not really conform to the image boundary (see attached picture).

Is there a way to enforce that the final mesh of tetrahedra entirely fills the image domain (as all labels are >0)? This is of particular importance for me since I have to apply non-trivial boundary conditions on the FE simulation that comes next.
<snip>

Hi Adrien,

Trying to achieve a similar aim using CGAL, though not specifically with the outer boundary, the best I have reached so far is using Mesh_criteria and a class "Proximity_domain_field_3" that provides a target characteristic length for a point (using operator()):
 FT operator()(const Point_3& q, const int dim, const Index& index) const

This value reduces as you close in on important regions. I then supply this object as the cell_size argument for the CGAL::Mesh_criteria_3 constructor, which in turn gets passed to CGAL::make_mesh_3. If your domain is cuboid, writing an efficient function to give a value would be pretty easy, but otherwise I have found AABB tree distance calculations works adequately, but only with some sort of discretized caching.

Not ideal, but if somebody can respond with an efficient way to constrain surface nodes during meshing, I will also be very happy :)

All the best,
Phil



Archive powered by MHonArc 2.6.18.

Top of Page