CGAL users discussion list

[Sebastien Loriot <>]

The way the closest primitive is found is the following:
using a hint point on one triangle, a sphere centered at the query point
and passing through that point is constructed. Then the intersection
of that sphere with the tree is computed. Each time the intersection
with a triangle is found, the radius of the sphere is decreased by
picking a point on the intersected primitive. In the end, a single
primitive is found.

I guess you could modify the procedure and do some kind of binary search
of the radius until the number of intersected primitives is the one 
requested (with some caching of radius per visited primitives for example).

Best,

Sebastien.



On 4/3/23 15:27, Martin Genet wrote:
> Dear all,
>
> For a project using mainly CGAL, I'm using an AABB tree to associate 
> each point of a point cloud to the closest triangle of a mesh using 
> closest_point_and_primitive(), but I would like to get the N closest 
> triangles (say 3 for instance).
>
> The AABB tree structure does not seem to provide any other method than 
> cleaning the tree and rebuilding it each time without the (N-i) th 
> closest triangle for 0 < i < N, which sounds unpracticable to do for 
> each point of a large point cloud.
>
> Am I missing something, or is there other solutions in CGAL that are 
> more suited for this purpose ?
>
> Best,
>
> Martin Genet
>
> --
>
> *Martin GENET*
> *Ingénieur d'Études, CNRS*
>
>
>
>
> *LIRIS UCBL1 Bâtiment Nautibus*
> 25-23 Avenue Pierre de Coubertin
> La Doua
> 69622 Villeurbanne cedex
>
>
> --
> You are currently subscribed to cgal-discuss.
> To unsubscribe or access the archives, go to
> https://sympa.inria.fr/sympa/info/cgal-discuss