CloudCompare forum

Hi there!

I am a rookie user, but CC is really powerful and helpful.
Currently we are dealing with the settlement or subsidence of a road surface. We have the point clouds of the road surface for two years. And I calculated the C2C distance and obtained the subsidence distribution over the road surface, which is satisfying. But it will be better if we can calcuate subsidence volume.

I don't know whether there is a specific function in CC to calculate the volume. But here are two thoughts of mine:
1) The appoximate mean distance calculated by chamfer distance via octree, multiplied by the road surface area, will be the volume
I don't have much confidence in this way, as the approximate mean distance should have a unsatifying accuracy. Right?

2) It might be possible to achieve the volume by integration of the histogram of the subsidence distribution.
All I need to do is to convert its "Y-axis Count" into area. It can be done by subsampling the points by Space or Octree, making the distance between points or the desity constant. Then I could probably calculate the area taken up by each point. By the way, which method of subsampling, Space of Octree, provide the most constant desite?
Then I need to export the histogram chart. With enough data of this histogram, it would not be hard to rebuild a histogram and do a integration in other mathmatic softwares. But is it possible to export the data of the histogram?

Or is there any better solution to this volume calculation problem? Please let me know.
Thank you very much!

Hi,

There's no tool to compute the volume right now (apart for a closed mesh but it's generally too hard to get one from a cloud comparison result).

1) As you guessed, the Chamfer distance is much too approximate.
2) This kind of calculus is much better I think. You can see an equivalent discussion with the M3C2 plugin: http://www.danielgm.net/cc/forum/viewtopic.php?t=755. The M3C2 plugin as the big advantage that it can compute signed distances between point clouds, and you can also force it to compute 'vertical' distances (which helps a lot when trying to approximate the volume).

And to get the list of distance values, you can simply export the compared cloud as an ASCII file. One of the columns will contain the distances. And it's easy to import this kind of files in Excel (you can use a semi-colon separator to get a true 'CSV' file) or in Matlab for instance.

Thanks a lot.

That's really helpful!

Hi,

I thought about how to measure the volume difference between two point clouds through measuring the volume of mesh.

I tried by placing two point clouds, one vertically above the other. Merge them and mesh them. When I use mesh(best fitting plane), it does not work so well.
But when I use mesh(xy plane), it kind of generated a good cylinder thing, judging from the outlooking.
Then I calculated the volume of the mesh, it indicated that the mesh has holes and the volume result is 3 times bigger the the result of the other reliable method.

Do you think this method can work or not?

thanks in advance.

I don't think you can trust the result of the '2D1/2 Delaunay' triangulation method for volume calculation.

Especially because it is only meant to produce a 2D1/2 mesh over a 2D1/2 cloud (and you obviously have a 3D cloud + you are looking for a 3D mesh). Moreover, the presence of holes in the mesh (whatever its type) invalidates the computed volume.

The only way to get a proper 3D mesh is to use the qPoissonRecon plugin. But it might not be very simple to get a proper mesh in your case. If you want to try you'll have to generate (good) normals on your clouds. But the lowest points should have a normal inverted relatively to the upper points if you want the mesh to be properly closed...

Thanks a lot!