[vtkusers] Point Cloud To Surface

[Kenneth Sloan]

Your basic problem is polynomial explosion of the Delaunay3D computation.

Here's an idea.

a) divide up space into OVERLAPPING chunks.  Each chunk should have a core which belongs ONLY to
that chunk, plus a surrounding guard zone which overlaps with adjacent chunks.  

b) solve the iso-surface problem (using Delaunay3D) inside each chunk

c) keep the part of the iso-surface inside the UNshared core of each chunk.

Note that this all depends on your point cloud being somewhat dense and regular.  The idea is that the Delaunay triangulation at the edges of each chunk's core area should match the DTs at adjacent
core areas.  To the extent this is NOT true, you will have some stitching artifacts.

There are many details to work out - but these depend on your data.

--
Kenneth Sloan
KennethRSloan at gmail.com


On May 14, 2013, at 06:26 , Simon von Bibra <simfirehawk at gmx.de> wrote:

> Hello Community,
> 
> still no answer to this question. Are there any further ideas?
> 
> Thanks Simon
>> Hello Community,
>> 
>> maybe this question has been asked already, but I could not find a good
>> answer so far.
>> 
>> I have a cloud of points, where each point has a scalar value. Now I
>> want to have a iso-surface for a given scalar value.
>> I tried already following approaches:
>> 
>> 1. Create a mesh with Delaunay3D:
>>       Works maybe ok for sizes of up to 10000 points, but for more
>> points it is unusable
>> 2.  Gaussian Splatter, but here I could not get it right. In the end I
>> just had some sphere, but no real iso-surface.
>> 
>> I thought, I can maybe regularize the point cloud with something like
>> marching cubes, but I did not get it to work. Also I will loose
>> correctness, because of the regular sampling.
>> 
>> Any ideas what will be a good way, to create the correct iso-surface?
>> 
>> Thanks in advance
>> Simon
> 
> _______________________________________________
> Powered by www.kitware.com
> 
> Visit other Kitware open-source projects at http://www.kitware.com/opensource/opensource.html
> 
> Please keep messages on-topic and check the VTK FAQ at: http://www.vtk.org/Wiki/VTK_FAQ
> 
> Follow this link to subscribe/unsubscribe:
> http://www.vtk.org/mailman/listinfo/vtkusers