Add ground filtering plugin (CSF) to the official repository of cloudcompare

[Haowyr]

Here is a little table for comparison between the latest release AND the experimental version I found and posted on the forum

----------------------------------------------------------------------------------------------------------------------
| Parameters | Experimental version(min:sec) | Latest Release(min:sec) | t2/t1 |
----------------------------------------------------------------------------------------------------------------------
| Relief + Slope processing | 1:00 | 2:40 | 2.7 |
| 2.0 / 500 / 0.5 | | | |
----------------------------------------------------------------------------------------------------------------------
| Relief + Slope processing | 2:50 | 21:30 | 7.6 |
| 1.0 / 500 / 0.2 | | | |
-----------------------------------------------------------------------------------------------------------------------
| Steep slope + Slope processing | 0:42 | 2:37 | 3.7 |
| 2.0 / 500 / 0.5 | | | |
----------------------------------------------------------------------------------------------------------------------
| Steep slope + Slope processing | 2:10 | 20:30 | 9.5 |
| 1.0 / 500 / 0.2 | | | |
----------------------------------------------------------------------------------------------------------------------

Time is roughly manually measured. Next is the screenshot for the last test of the console with the latest release.

[daniel]

Ah! There's obviously something wrong with the 'Rasterize' step ;).

Can you maybe send your cloud to me? I guess it has someting to do with its configuration...

[Haowyr]

Exactly what I noticed too. I'm looking for a way to upload the point cloud and send it to you ;)

EDIT: Done by email!

[daniel]

I haven't received it. Can you try: daniel.girardeau [at] gmail.com ?

[Haowyr]

Yup, my mail failed and I didn't realized before this morning, it should be now in you inbox at 'cloudcompare@danielgm.net' but I send it on your other email adress ;)

[wpqjbzwm]

@Haowyr Are yo applying CSF to terrestrial Lidar data? How many points in your point cloud. Usually, terrestrial Lidar data contains too many points, which may slow the rasterization process, since it needs to find the nearest points. Another problem is the setting of cloth resolution. When it is increased from 2 to 1, the cloth particle will be increased by a factor of 4.

[Haowyr]

Yes I am, but on a parsed terrestrial cloud, the cloud I sent Daniel is just under 1 million points, it actually works pretty correctly. I totally get the *4 factor when I'm changing the size of the cloth resolution.
What I was pointing was the difference between the two versions of the plug in, which, at least in my mind, shouldn't be worse in the official release, maybe even faster as Daniel explained he deleted some useless steps in the script.

[daniel]

It's probably a bug due to the modifications I made to 'increase' the speed :D. I'll fix this asap.

[Haowyr]

Great to read this! Let us know when you're done for some more testing ;)

[daniel]

In fact I overlooked the possibility that the cloud could be quite far from the initial (cloth) grid. This was resulting in very far neighbor search and my code was not able to handle this properly. Waiting for a better approach, I reactivated the old code (based on CGAL) for the 'rasterize' part.

I updated the latest beta release:
http://www.cloudcompare.org/release/Clo ... bin_x64.7z

The final part (cloud / final grid distance) still uses CC's octree and is therefore much faster.