radius

[vincent.compain]

I begin to use the software and I try to compare two TLS cloud points but there are several things I don't understand.
I don't understand how to compute the "radius" for the density, the normal to compare my cloud points.
Can you help me?
Thank you
vincent

[daniel]

The density is a kind of local statistics, therefore you have to define how 'local' it should be. In effect you also have to find the radius yourself as it depends on whether the cloud is uniformly sampled or not, etc. Generally, the default value is good (it should be a percentage of the cloud bounding-box, but don't hesitate to make some trial and errors ;).

And for the normals, you have the 'auto' button in the 'Edit > Normals > Compute' tool that can help you. But once again it depends on your cloud so you may have to adjust your parameter. And by the way, the normals are only needed by the M3C2 algorithm.

[vincent.compain]

Thank you Daniel
Yes in fact I try to work with the plugin M3C2, it's for that que I try to understand the radius, normals, density, etc.
I'm working with two cloud points that I subsampled with a scale of 0.05 m (cloud #1 : 1350 000 points and cloud #2 : 1 350 000 points).
After that, I calculate the normals with Normals > compute then remove the bad points. To finish, I started the plugin M3C2 with these parameters :
scales : normals = use cloud #1 normals ; projection = 0.3 ; max depth 1; core points = use cloud #1 and registration error = 0.02

the result seems good but as I don't underwell very well, it's very subjective

thank you for your help

Vincent

[daniel]

Have you read the article? Otherwise if you use values that are close to what the 'guess params' returns, then it shouldn't be that bad.