[pdal] About litree filter output: "PDAL: All points collinear"

[Jim Klassen]

On 12/16/22 19:19, Ulises Ibarra wrote:
> Hello everyone:
> I have a question about the use of filters.litree. When I run:
>
> $ pdal pipeline ./litree_prueba.json
>
> where de ./litree_prueba.json:
>
> [
>        "./input.las"
>     {
>         "type":"filters.hag_delaunay"
>     },
>     {
>         "type":"filters.sort",
>         "dimension":"HeightAboveGround",
>         "order":"DESC"
>     },
>     {
>         "type":"filters.litree",
>         "min_points":50,
>         "min_height":3.0,
>         "radius":50.0
>     },
>     {
>         "type":"writers.las",
>         "filename":"./output.as <http://output.as>",
>         "minor_version":4,
>         "extra_dims":"all"
>     }
> ]
>
> I get de following output:
> "PDAL: All points collinear"
>
> And the file "output.las" was never created.
>
> Could you guide me on what I'm doing wrong, please?
>
> Thank you
>
> Ulises
>

I've seen that error before when using hag_delaunay.  It's complaining it can't make triangles (delaunay triangulation) because it thinks all of the input points fall on a line.  I found that sometimes hag_nn will work in these situations.  It has been awhile since I've hit that error (so I was using an older version of PDAL), it was sporadic (only occurred on some input tiles) and I didn't have time to track down the exact points that were causing it.

I'm assuming (dangerous, I know) that PDAL is looking at some subset of close points when it is constructing the triangulation and that the error is stating that all points were collinear in one/some of those subsets rather than saying that the entire point cloud is collinear.  However, if all of the source points are actually all collinear, then that would explain it.

If I remember right, this tended to happen on messy (noisy), very dense (higher point density than reasonable given the precision of the measurements) point clouds where I suspect that it may have been possible for the point cloud to have areas where there were many points at the same or nearly the same X,Y,Z.  Another possibility was the error being triggered by some odd (noise) points forming nonsensical geometries.  But as I said, I never did have time to track down the root cause for sure.
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.osgeo.org/pipermail/pdal/attachments/20221221/89988db7/attachment.htm>