Steep Cliff Face

[athum]

 Hi all, I'm working on a project which requires classification of a cliff/rock face that has scattered vegetation across it. The ledges have varying slope angles across the cliff. Might anyone have any suggestions for parameters or specific tools to use that could render  good ground classification across the model. Thus far I've had decent luck getting the cliff face classified as ground and the vegetation across as high veg. However, there is still a lot of cliff face (rock) classified as high vegetation and unclassified. Please let me know if anyone has found a good workflow for this type of classification. Thanks.

Current parameters for lasground_new:

 lasground_new -v -i %TEMP_FILES%\denoised\*.laz -feet -elevation_feet -wilderness -not_airborne -ultra_fine -step 1.5 -bulge 1.5 -offset 0.1 -spike 0.50 -compute_height -odir %TEMP_FILES%\tiles_ground_town -olaz -cores %NUM_CORES%

[Support at rapidlasso]

Hi Austen,

thanks for posting this to the group.

As you already saw there was a quite similar question some time ago:

https://groups.google.com/g/lastools/c/6O-qzDVHBwo/m/ph2P4zrkDAAJ

Can you provide some small samples so we can try to solve this?

Thanks,

Jochen

[Jochen Bind]

Not to step outside of the LAStools realm here, but I think using multiple software packages to achieve a goal is often quite successful. This case, to me, represents a real 3D problem, rather than a 2.5D problem. LAStools’ approach to ground classification basically works in 2.5D.

We came across similar issues when trying to classify vegetation along near-vertical river banks. The approach we took was using the CloudCompare implementation Canupo (based on the paper available here: https://nicolas.brodu.net/common/recherche/publications/canupo.pdf) and got very good results.

 

Jo

--
Download LAStools at
https://rapidlasso.de
Manage your settings at
https://groups.google.com/g/lastools/membership
---
You received this message because you are subscribed to the Google Groups "LAStools - efficient tools for LiDAR processing" group.
To unsubscribe from this group and stop receiving emails from it, send an email to lastools+u...@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/lastools/6bd0a083-4ef7-40c9-a18b-46451f57ba4cn%40googlegroups.com.

Jochen Bind Hydrodynamics and GIS Technician +64-3-343-8021 +64-21-087-54696 National Institute of Water & Atmospheric Research Ltd (NIWA) 10 Kyle Street Riccarton Christchurch New Zealand Connect with NIWA: niwa.co.nz Facebook LinkedIn Twitter Instagram

To ensure compliance with legal requirements and to maintain cyber security standards, NIWA's IT systems are subject to ongoing monitoring, activity logging and auditing. This monitoring and auditing service may be provided by third parties. Such third parties can access information transmitted to, processed by and stored on NIWA's IT systems

[Jorge Delgado García]

The truth is that automatic classifications are automatic, so it is not easy to get the results we want (at least not without further editing), especially in the case of complex cases. In any case, what is necessary is to analyze what can allow the system to differentiate between the different classes, both geometrically (roughness, distance to the plane, etc.) and, very importantly, radiometrically.
If in the end it is complicated to apply geometric elements, and as long as your point cloud has an associated RGB, or even with the reflectivity itself - let's not forget that most sensors work in IR - it is usually more or less simple to calculate a "vegetation index" (or an approximation to it) and discriminate what is "vegetation" from what is bare rock or soil. Do you consider it viable to apply this methodology?
Another possibility is not to try to perform the classification in one step, but to perform the classification in several steps, in order to be able to better adjust the parameters in each case. For that, LAStools is a great tool, you can use the -keep_classification and -drop_classification commands to select in which zones you want to reclassify, override the classifications and reclassify again.

As for the CCompare methodology you propose, it obviously works with distances to the plane normal to the surfaces, it is possible that it will give you good results, as long as you can define the planes well. On the other hand, in terms of calculation it is very very intensive.

Jorge Delgado

jdel...@ujaen.es

To view this discussion on the web visit https://groups.google.com/d/msgid/lastools/SY4PR01MB7106E3E6D93159464291217FD96E9%40SY4PR01MB7106.ausprd01.prod.outlook.com.

--

[Jochen Bind]

Hi Jorge,

You raise some good points about making use of the IR information in the intensity of returns. In our case, we’re using a Velodyne scanner and have no RGB information. We did preliminary filtering out of water based on intensity.

 

You are correct in that Canupo requires some processing, but we’ve processed point clouds of about 150 Mio points on a 5-year old desktop without major issues. A small section of the vegetation classification results are shown in the attached image.

 

Cheers,

Jo

To view this discussion on the web visit https://groups.google.com/d/msgid/lastools/CADgtU__DXO-HsnQNf72ywud51Sib5MFWf3f3hf1FZ3RYbnkGSg%40mail.gmail.com.