[vtkusers] Re: Fill contours with Pixels

[Mike Jackson]

On Feb 14, 2007, at 3:45 AM, Hendrik Belitz wrote:

> Hi Mike,
>
> Mike Jackson wrote:
>> I have a 3D triangle mesh that I am trying to close the "Holes"
>> on. This mesh is pretty complex and although I have searched the
>> archives I have not found anything that has worked out. The main
>> problem being that a "contour" line can traverse 2 or more planes.
>
> I must admit that I didn't understood what you're expecting from  
> the algorithm you're searching for. This is mainly due to the fact  
> that I don't know what you mean by the terms "hole" and "contour".  
> Is this about closing topological inconsistencies in your original  
> mesh? Or is it about closing surface areas which are fully  
> delineated by surrounding edges, but (for some strange reason)  
> without any kind of face lying in this area? Or did I get you  
> totally wrong and you're speaking about something completely  
> different?

Nope, this is what is happening. They say a picture is worth a  
thousand words so here is a picture of the mesh as it stands now:

http://www.bluequartz.net/trianglemesh

I have two sided lighting turned OFF in this image so the black  
colored areas are actually the "back" of triangle faces, which are on  
the interior of the mesh. The goal is to fill these "holes" with  
triangles in order to make the mesh water tight.

The mesh was done by another person and the code that they used  
didn't fill the holes. I do not know why.

>
>>    So what I was thinking was to extract the contour loops just at
>> the boundary surface. This gives nice loops. Now if I could
>> somehow "fill" those loops with solid pixels, convert the whole
>> thing to a voxel volume, I believe I have a way to "extract
>> surface" of the new voxel volume, triangulate it, than append and
>> merge to the original mesh.
>
> Please note that by voxelizing your geometry you introduce a  
> discretization error depending on your sampling grid. The surface  
> extraction with high probability will also introduce new  
> topological errors in your final result. Is this really no problem  
> for your application?
>
> Regards
>  Hendrik
>

I did not realize this could happen so it is good to know.

Thanks
-- 
Mike Jackson   Senior Research Engineer
Innovative Management & Technology Services