Vertical Surfacing

[AndrewRoberts]

Hey guys,

Just working with some ground based LiDAR datasets of small coastal communities. When I mesh my point clouds they seem to mesh very nicely with respect to the x,y axis but not to the z axis. It appears it triangulates the points with other points with the closest x,y coordinates but doesn't take the z axis into account.

This can be seen in the image from the wiki help...


but also in my dataset as well...



I can see this being useful with airborne LiDAR datasets but with my ground based unit I tent to have more hits on vertical structures than an airborne LiDAR would have. Let me know if anyone has a fix for this!

Thanks again!
-Andrew

[daniel]

Indeed the 2D1/2 Delaunay triangulation is only meant to be used with clouds having no point on top of others (this is what we mean by "2D1/2" - i.e. not truly "3D").

To mesh real 3D point clouds, CC is not always the best solution. You can try the 'Poisson Reconstruction' plugin (qPoissonRecon). But you'll need good normals to use it. And this algorithm always creates a 'closed' mesh, so you generally have to clean the resulting mesh afterwards.

[AndrewRoberts]

Thanks Daniel. I was wondering about this!

I was at a demontration for a software (I believe it was hypack) where I believe they did two 2D meshes. One from the top down (like CC) and another from a side angle. It then, I assume, somehow merged the two resulting surfaces, picking the smoothest surfaces as the correct surface. It was pretty interesting.

-Andrew