CGAL users discussion list

["Sebastien Loriot (GeometryFactory)" <>]

On 03/11/2016 08:44 PM, Sukhraj Singh wrote:
> Hello,
>
> "Triangulated Surface Mesh Segmentation" uses SDF values to do the
> segmentation of a surface mesh.
> I wanted to used the SDF values for matching. As described in
> http://research.microsoft.com/en-us/um/people/rgal/publications/papers/pose_oblivious.pdf.
> This paper introduced the idea of SDF.
>
> The CGAL implementation of SDF values digresses somewhat from what is
> mentioned in the above research article.
> One of the digression is that an SDF value is calculated at the center
> of a "face"  instead of at a "vertex" . That is okay.
> But to calculate the signatures the SDF value is multiplied by a weight
> which is influence of that vertex. Where influence is the (area of all
> the faces connected to that vertex)/(total mesh surface area). Later a
> 64 bin Histogram is created form these weighted values.
> My question is,since in CGAL implementation "SDF" is calculated at the
> Face center what would be the best way to calculate the
> weights/influence in such a case.
>
> Thanks to comment.

It don't think it is hard to adapt the CGAL code to use
projection from a vertex instead of from a face centroid
(it is just a matter of changing the set of faces to
ignored while shooting the ray).

Sebastien.