Export classified point clouds

[daniel]

Oh, then you should now use the 'Edit > Scalar fields > Filter by value' (with the right range, and with the 'Classification' scalar field active).

[xabierr]

Hi Daniel,

What would be the best approach to manually reclassify classes? For example, all 2 "ground to 3 "low vegetation"? Is "Scalar fields arithmetics" the best way?

Thanks

Javier

[daniel]

Ah that's an idea, but I believe you can now use the new 'CloudLayers' plugin to do that interactively?

See https://www.cloudcompare.org/doc/wiki/i ... s_(plugin)

[apa3495]

Hello CC users,
Currently trying to classify manually a pointcloud with custom classes (using Cloud Layer).
The LAS export did seems to preserve the scalar field (renaming the scalar field with "Classification" worked well) but the classes information are lost (replaced with the conventional ones, ground, vegetation etc).
Anyboby knows how to tackle this issue ?
The lack of documentation of the Cloud Layer plug-in does'nt help.

Thanks for the support

[daniel]

All classes are lost? Or only the ones above a certain value?

[apa3495]

Hello Daniel,

Apparently, the codes (i.e. index/value of the classes) seems to be preserved but the label (classes names are not).
This is valid only if I talk about export and reimport in las, with Cloud Clompare.
(I have other issues that seems more related to potree converter/viewer).
It appears that the classes are more or less preserved if you re-open the las on the same computer but are lost while opening the .las on other computer. I tested with standard classes (0-12) juste changing the labels, then tried on user defined classes range between (64 an 76).

In the first case, my classes were lost (replaced with default one).
In the second case, all labels were lost but also some classes.
So to answer your question, it depends of the classes used.

Looking at the code of the plug-in on Github it seems that the plug-in could work only with classes standard or user definable already loaded by the plug-in.

I don't know if it's clear, but it seems that the plug-in doesn't handle well the read and write of custom classes.

Thanks for your help

[daniel]

So the labels are indeed not something stored in the internal cloud structure of CloudCompare, nor in LAS files (there's no way to save it there). It's only in the plugin for convenience.

[dan_uhl_midi]

Hi All, following this thread. My question is related but different. Is there any way to export label information, such as coordinate?

It seems that it is impossible, even, to copy/paste the label data fields such as coordinates. My workflow is below, and I'm wondering if anyone has an idea how I could make this any less manual (automate/code). One thought that comes to mind, is (when you read below and see the workflow), I could export each of the tiny rockfall selections in one layer.

The project is:
I am identifying, based on an M3C2 layer scalar values, rockfall events on a large area ( 2 x 1.5 km wall). So far over 600 events. I have manually added labels to each event. I have 5 identical clouds, each with their own scalar value:

1)-Original cloud, time = beginning of period (time = 0)
2)-Original cloud, time = end of period (time = 1)
3)-Cloud with Aspect as scalar value projected on it (thanks to 'interpolate from another entity' tool which is great)
4)-Cloud with Slope as scalar value projected on it (thanks to 'interpolate from another entity' tool which is great)
5)-Cloud with M3C2 values representing change areas.

For EACH discrete rockfall (+600), I select all layers and make a cut around it, and place the selections in their own group corresponding to the label number.

Then, I'm generally using 'Compute 2.5D volume' after rotating the rockfall so it is perfectly oriented on one axes (in this case Y but could either X,Y,Z)

Then using 'Compute stat params' I get the gaussian mean of the slope, aspect, and M3C2 values for the analysis.

All data is entered manually in Excel.

Aside from the original question of if there is any way to extract the info in the label (sounds like not), does someone have a suggestion for making this workflow better? At this point it will be almost 80 hours to get through all of it with my current workflow.

Thanks!

[daniel]

At least the workflow is well thought out ;). I don't see anything that could make it much more efficient... Of course you may be able to automate parts of it (especially the part before the manual 'leveling' of each discrete rockfall, and maybe the part after it, but that would be more challenging...). For the first part, the command line mode might be an option (especially thanks to some latest additions in v2.13.alpha) otherwise the CloudCompy python interface might also be an option (it generally gives a better / lower level access to CC's functions).

For labels, the only thing you can do right now is indeed to copy/paste the 'Body' field in the label properties... And I'm not sure having a way to export a single label, or even all the labels of a single cloud to a file would help that much?