CGAL users discussion list

[Juliano Costa <>]

Hello Manuel,

 

Some additional issues:

 

1. About Alpha Shapes:

- As you suggested I tried to do something nested in order to set my ->info() information along with the added vertices.

My used declaration is like this:

 

typedef CGAL::Exact_predicates_inexact_constructions_kernel K;

//My info is just a ordered index for the vertices

typedef CGAL::Triangulation_vertex_base_with_info_3<CPUInt64_t, K> Vbi;

typedef CGAL::Alpha_shape_vertex_base_3<K,Vbi> AsVbi;

typedef CGAL::Alpha_shape_cell_base_3<K> AsCb;

typedef CGAL::Alpha_shape_3<DelaunayTriangleHierarchy> Alpha_shape_3;

typedef Alpha_shape_3::Alpha_iterator Alpha_iterator;

typedef Alpha_shape_3::NT NT;

 

Then I got these errors:

error C2039: 'set_down' : is not a member of 'CGAL::Alpha_shape_vertex_base_3<Gt,Vb>'

error C2039: 'set_up' : is not a member of 'CGAL::Alpha_shape_vertex_base_3<Gt,Vb>'

Am I doing anything wrong about your suggestion?!

 

I didn't understang why if I use "incident_cells(infinite_vertex(), output_iterator)" I wouln't

need the index any longer.

 

I'm using this ->info to set the index, so after the triangulation the indexes of the vertices

are the same as when I add it, since after the insertion the initial order is not guaranteed,

right?! So I guess this info is fundamental. Or not???

 

2. About the Surface Reconstruction:

- The Surface Reconstruction seems to be great for any kind of points cloud but

I have a point cloud with x,y,z only (not the normals) and then I tried to use the

example  "..\examples\Point_set_processing_3\normal_estimation" to calculate

the necessary oriented normals for the Poisson reconstruction.

 

I'm trying to use it on a x,y,z file of a points cloud with 50000 points and the

program is freezing on the second fase (normals orientation). See the output

of the program:

 

tunnel_50mil.xyz tunnel_50mil_normals.xyz
Normal estimation
Open tunnel_50mil.xyz for reading...
Reads file tunnel_50mil.xyz: 50000 points, 2.496 seconds
Estimates Normals Direction by Jet Fitting (k=18)...
Calls jet_estimate_normals()
  6 Mb allocated
  Creates KD-tree
  11 Mb allocated
  Computes normals
  14 Mb allocated
End of jet_estimate_normals()
done: 52.405 seconds, 7 Mb allocated
Orients Normals with a Minimum Spanning Tree (k=18)...
Calls mst_orient_normals()
  7 Mb allocated
  Create Index_property_map
  Index_property_map: optimized version for a random access container
  mst_find_source()
  7 Mb allocated
  Creates KD-tree
  13 Mb allocated
  Creates Riemannian Graph
//From here (...) Never finishes!

 

Thanks a lot for all your help! 

 

Juliano Costa
 

2009/8/5 Manuel Caroli <>

Hi Juliano,

Juliano Costa wrote:

Hello again Manuel,
 Yes. By "surface" I was refearring to the convex hull (the outter side of the envelop points).

Note that you can use

incident_cells(infinite_vertex(), output_iterator) to get the cells that have the triangles of the convex hull as a facet.

 This new package for "Surface reconstruction" seems just GREAT.
It looks exactlly what I need to do.
I'm going to download the beta version to test it now!
 Just one more thing:
I was trying to to integrate the alpha shape with my current example.
Right now I have CGAL::Triangulation_vertex_base_with_info_3 in order to set the ->info() with my
index.
 For the alpha shape, If I really understood it, I must use CGAL::Alpha_shape_vertex_base_3 but this does not contain
the "info" field.
Is there any other way for me to set the "info"?
There should be an CGAL::Alpha_shape_vertex_base_with_info_3 file to do so, right?!

It should be possible to use them in a nested way, i.e.
Alpha_shape_vertex_base_3<Traits,Triangulation_vertex_base_with_info_3>

Maybe with the above mentioned approach to extract the convex hull you don't need the info anymore...

best

Manuel