songsenand
/
cgal
mirror of https://github.com/CGAL/cgal
1
0
Fork 0
Code Issues Packages Projects Releases Wiki Activity

cosmetic change in the doc

This commit is contained in:
Sébastien Loriot 2014-10-27 16:05:08 +01:00
parent a801ef2728
commit d1c65b4986
1 changed files with 1 additions and 1 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

2
Scale_space_reconstruction_3/doc/Scale_space_reconstruction_3/Scale_space_reconstruction_3.txt
View File

@ -87,7 +87,7 @@ The main class `Scale_space_surface_reconstruction_3` contains all the functiona
The neighborhood size is estimated using `Orthogonal_k_neighbor_search`. The point set is generally stored in a `Orthogonal_k_neighbor_search::Tree`. When the neighborhood size is estimated, this tree is searched for nearest neighbors.
The scale-space is constructed at the original scale of the points. An iteration of increasing the the scale is computed using a weighted PCA procedure. As described by Digne <i>et al.</i> \cgalCite{cgal:dmsl-ssmrp-11}, unlike similar methods this procedure does not lead to an "undesirable clustering effect". By default we use the efficient \ref thirdpartyEigen libraries for this procedure. It is also possible to provide your own model for the `WeightedApproximation_3` concept and provide it as a template parameter. The PCA procedure is performed locally per point, so it can performed with parallel computing if both Intel TBB is linked and the `Ct` template parameter is set to `Parallel_tag`.
The scale-space is constructed at the original scale of the points. An iteration of increasing the scale is computed using a weighted PCA procedure. As described by Digne <i>et al.</i> \cgalCite{cgal:dmsl-ssmrp-11}, unlike similar methods this procedure does not lead to an <em>undesirable clustering effect</em>. By default the efficient \ref thirdpartyEigen library is used for this procedure. It is also possible to provide your own model for the `WeightedApproximation_3` concept. The weighted PCA procedure is performed locally per point, so it can be performed with parallel computing (linking with Intel TBB and passing the `Parallel_tag` to the reconstruction class is required).
The mesh reconstruction is performed by filtering a \ref Chapter_3D_Alpha_Shapes "3D alpha shape" of the point set at a fixed scale. This filtering constructs a triangle for each regular facet; each singular facet results in two triangles facing opposite directions.