Different Results Cloud-to-cloud distance

[yvdnbroe]

Dear all,

I have a question about the cloud-to-cloud distance tool. Depending on which cloud I select as reference role/compared role, I get different results in mean distance. Nevertheless I use two exact same clouds when I which between roles.

How is this possible? When the clouds are the same, the difference between both must be the same, not?

Kindly regards,

Yannick Vandenbroeck

[daniel]

In fact the comparison process is not symmetrical.

This is mostly due to the fact that:

• we are using the distance to the nearest neighbor (and depending on the cloud densities and sampling pattern two points are rarely 'mutual' nearest neighbors - it's hard to explain but with a little drawing it's really easy to understand ;).
• another big difference is generally the number of points
• and the holes or other non overlapping areas can also cause some big differences: if you compute the distance of points in an area where there are no points in the other cloud then you'll get big distance. But if you compute the distances of the other cloud you won't get anything (as there are no points in thus same area!) and therefore you would get lower distances in average.

All in all this is why you have to choose the 'reference' cloud. And you measure the displacements of one cloud 'relatively' to the other.

[yvdnbroe]

Thank you very much for your fast and clear answer.

Can you just tell me which formula CC uses to calculate the mean distance and std. deviation?

I tried to calculate it manually by exporting the histogram, but I don't get the same values as calculated by CC.

Thanks in advance.

[daniel]

We use the very standard formulas (see wikipedia) but you can't compute them from the histogram. You need the whole set of distances (i.e. one distance per point). You can get them by saving the cloud as an ASCII file (you'll get one point per line, with the distance as 4th column - or more if the cloud also has colors or normals for instance).

You can load ASCII files in Excel for instance (if there are not too many points).