[pdal] LAS Class Flags (Withheld, Synthetic, Ignored) in PDAL

[Charles Karney]

My own first-hand experience to the mask flags was to insert synthetic
ground points on a grid wherever real ground points are absent.  This
was a requirement of our client.  This means that the ground points
(real + synthetic) cover the entire tile.

Some of the data provided by NGA has been processed in this way and so
includes synthetic ground points.

If the resulting dataset (including synthetic points) is reprocessed for
whatever reason, the synthetic points should normally *not* be part of
the analysis.  Our example our rule for detecting building points starts
by asserting that there shouldn't be ground under any building points.
If we repeated the ground extract with the synthetic points, there would
be ground everywhere and so we would have no buildings.

The choices are then:

* discard the synthetic points;

* ignore the synthetic points in the processing but retain them in the
   dataset.

The first approach is easy with PDAL (apply a range filter first).  The
second approach works only for some filters and may require repeated
applications of an "ignore" option for each filter.  It would be "nice"
if this diversion of classes of points around a filter chain were
easier.  However, I recognize that this is significant architectural
change.

I can imagine that similar considerations apply for the withheld and
ignored flags.  But I haven't encountered datasets with these flags set.


On 6/12/20 10:31 AM, Howard Butler wrote:
> All,
> 
> We are currently discussing some kind of refactor or enhancement to the handling of these kind of mask flags in PDAL, but we are wondering how much impact these changes would actually have. For the most part, these things have been "just data" and PDAL users were on their own to keep track of masks and preserve them and so on. That might mean that people have lots of custom bits to manage them, but we don't know.
> 
> Are you managing LAS-style Classification flags yourself with PDAL? If so, can you tell us what you're doing now and how you might like to see more regimented facility in PDAL (if at all) for dealing with such things? Some feedback would help guide us on how extensive of a refactor we could feel comfortable taking on the topic.
> 
> Howard
> _______________________________________________
> pdal mailing list
> pdal at lists.osgeo.org
> https://lists.osgeo.org/mailman/listinfo/pdal
>