AnsweredAssumed Answered

How does spatial sampling work for point clouds?

Question asked by James Maeding on Sep 7, 2018
Latest reply on Sep 10, 2018 by James Maeding

I have used many point cloud toolsets, and am now using TBC. One of the main tasks we do is make a TIN surface from ground classified point cloud region. Of course, if you use all points you get too many triangles, and that process of thinning down the points, but not smoothing the surface too much is a big deal when you have curbs and hardscape involved.

So in TBC, the only options of thinning down points is by sampling. The two options are not clear to me though.

I am wondering for the spatial sampling: Is the sampled point the average of all the points within the sample area size? Or is it a point taken in the middle of the area based on immediately surrounding points?

Also, does the random sampling method work the same?

Really, this kind of sampling is not good enough for the task at hand, as it does not distinguish between "important" points, and those that form a slightly varying plane. Civil's want the best and smallest set of triangles to describe the ground.

We want big triangles for flat areas, and little tris for areas that vary a bunch.

TBC is missing this kind of analysis, so I normally use the Carlson pt cloud tools to make the TIN. So workflow is TBC to classify and clean the ground region, and Carlson to make the "optimized" TIN. Its not great though. The Carlson tool needs to make tris smaller in some areas, and larger in others, but at least it does an ok job.

Anyway, I wanted to understand the sampling as it has to at least do an averaging or its not that useful to me.