[pdal] Matlab filters and Python refactoring

[Howard Butler]

All,

Similar to PDAL's Python support https://www.pdal.io/python.html <https://www.pdal.io/python.html>, I have embarked on providing the ability to embed Matlab in the form of a filters.matlab stage. A client has requested this capability, and while I'm quite familiar with Python embedding scenarios, I hope the community can provide some guidance on what the must-haves would be for Matlab capability. The first attempt will simply allow users to touch a special "pdal" array with their script and update the point values as needed with no ability to change the schema in any way. The typical processing scenario would use filters.matlab in a pipeline operation. Assuming that is working in the upcoming 1.6 PDAL release, what other features would PDAL's Matlab connections need to make them extremely useful and powerful?

Additionally, as part of the Matlab effort, it's clear the PDAL Python stuff is spread about a bit too much. The filters.predicate and filters.programmable nomenclature is confusing, and I propose that we simply rename it all to filters.python, and if the returned dictionary of Numpy arrays has a "mask" array on it, it will behave like filters.predicate does to cull the points. Otherwise, not much else will change for PDAL's Python stuff besides the filter name.

Would Matlab interaction in PDAL pipelines be useful to you? What other PDAL/Matlab interaction scenarios give you the power of PDAL's processing pipelines married with Matlab's deep prototyping environment? My client would like to use intricate Matlab algorithms in the context of PDAL pipelines for data processing orchestration. How do we expose PDAL to the Matlab user in a way that would be useful for them? 

Howard
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.osgeo.org/pipermail/pdal/attachments/20170619/774a8866/attachment.html>