CloudCompare forum

Hi,

i'm currently working with the "Geometric Features" tool to evaluate the quality of spheres.
For test purposes I have generated some artificial spheres (with and without noise - https://imgur.com/a/WVluRme) with a radius of 38.1 mm with 40 000 points.
What would be a good local neighbourhood radius for this task? I can't even get close to a sphericity of 1.
Or am I better of fitting a sphere to the cloud and then evaluating the radius?

Thanks a lot four your time.

The 'sphericity' measure is described here: https://ethz.ch/content/dam/ethz/specia ... pr2016.pdf

It's very theoretical (the ratio between two eigenvalues of the covariance matrix of the local set of points :D). But from what I understand, it would give 1 if the local neighborhood was spherical! Not part of a sphere. Basically you would have 1 if the radius was big enough so as to encompass all the points...

So, probably, using the Sphere fitting tool and looking at the RMS would be better ;)