Voxelise data or import voxelised data

[Anne Baranger]

Dear DART team,

Thank you for all your work and for making it available to the scientific community.

I believe that the issue I'm facing may have been asked before, but I’m not sure I fully understood the previous answers. Here is my situation: I have a terrestrial LiDAR point cloud of a forest that has been classified into leaves and wood. I would like to voxelize it in order to run DART simulations. I am unsure whether the voxelization must be performed with pytools4dart, or if I can voxelize it myself and then load the voxelized data into DART. In the latter case, I did not understand how to proceed.

Do you have any advice?

Best regards,
Anne Baranger

[Tiangang Yin]

Dear Anne,

There are several ways you can voxelize terrestrial lidar data, depending on whether you just want a structural outline or a detailed 3D distribution of leaf area density (LAD). 

For the former method, it is quite easy: after 3D gridding, wherever there is a vegetation point, a voxel will be generated. You can empirically define the LAD distribution, maybe choosing an average value and assigning it to each voxel.

For the latter method, you would need to use an external method to generate the 3D LAD distribution by tracing the ray of each lidar pulse. Two examples are the VoxLAD model or the AMAPVox model.

https://www.sciencedirect.com/science/article/pii/S003442571730233X

https://www.sciencedirect.com/science/article/pii/S0168192311001481

https://www.sciencedirect.com/science/article/pii/S0168192313002608

https://amapvox.org/

Perhaps Florian can provide some insights because he is the major developer of pytools4dart

https://ieeexplore.ieee.org/abstract/document/10242262

Once you have the voxels, you will need a tool in DART called DARTDAO to import the voxels. The documentation of DARTDAO is under the directory "bin\python_script\DAO\doc" of your DART installation folder.

Additionally, our lab provides a script initially developed by Eric Chavanon. You can find the script from the following webpage:

https://www.3msi.net/foscenes 

The script has been adapted to the FoScenes dataset, which is generated from open-access airborne lidar data, but the importation is the same as TLS. You can download the data and check the format of the voxelized scene.

Lastly, I believe another way is that you could share the voxelized scene with us, and we will help you to generate the DART simulation and send it back to you.

Let me know how you would like to proceed. Florian may provide additional information. 

Best regards,

Tiangang

[Anne Baranger]

Dear Tiangang,
Thank you very much for your quick and detailed response. My situation does indeed correspond to the second case you mentioned, with a detailed 3D distribution of the LAD. I will look into the FoScenes example and get back to you if I have any further questions.

Best wishes,
Anne

[Anne Baranger]

Dear Tiangang,

I am trying to run the Foscenes examples to better understand how DARTDAO works. I am using this code: https://drive.google.com/file/d/1V--MaMmpS29-hh7uTy2641xrK19kz3v1/view?usp=drive_link

However, it is not working properly. I think this is because I did not load the DEM properly in the first steps mentioned in the readme :  
"1) simulate the basic simulation:

* create a DART simulation in your DART's simulation folder
* open the DART GUI editor and select "Forward" mode (in "Passive RS & RB" section)
* eventually tweak the number of thread(s) and the number of samples you want depending on your computer capabilities
* prepare a DEM txt data and import it into DART to generate a DEM.mpr data"

I am not sure how to complete this final step. When I tried to convert the dtm.txt file directly from the DART editor (in the "Earth scene" tab), it created a DEM.mpr file. However, when I loaded the file afterwards when running the python script, it was empty. What is the correct way to complete this step?

Best regards,

Anne

[Tiangang Yin]

Dear Anne,

I do not think the 1st three steps you mentioned will influence the last step.

Can you send us the DEM txt data so we can check what happened? 

BTW, the lidar tools in DART should work in the following release to directly generate voxels from points. Eric has corrected some bugs in the previous release. We will let you know when it is ready. Stay tuned. 

Best regards,

Tiangang

[Anne Baranger]

Dear Tiangang,

For the moment I am trying to run one of the Foscene examples : Northwood 
Following the readme of the associated python code, I tried to import the dtm.txt file, using the DEMgenerator in the Editor tab. Then I did not manage to run "Dart" module.

Instead I ran Direction, Phase, and Maket and then the python code (example_FT_leafvox.py). After this, I managed to display the mockup with voxelised leaves. However, the DTM did not load correctly and displayed a brown square that was separate from the leaves' voxels.

Do you know what step I am getting wrong?

Best regards,
Anne