C2C non-absolute distances

[hinkf]

Hello all,

I am comparing two clouds of the same model using C2C distance analysis.
Is there an option or possibility to compute the non-absolute Z-distances? I wand to investigate if one scan is tilted in respect to the other - to do so, I need to know if the Z-distances are positive or negative.

The M3C2 plugin seem to have this possibilty - but I'm not very comfortable with their normals and projection parameters.

Thanks a lot for your help,
Florian

[daniel]

He he, if you want signed distances then you need normals ;)

You have multiple choices:
[*] either compute a triangular mesh on your reference point cloud (with Delaunay or Poisson)
[*] or using M3C2 indeed (if you don't want to 'fight' with M3C2 options, you can compute the normals yourself on the reference cloud prior to calling the plugin - with Edit > Normals > Compute)

[olivier.martin]

Hello Daniel,

I re-opened the post because I have a similar question.
I do not understand why we need normals to have signed X, Y or Z-distances ? I understand for the global distance between the two points but why for the distance on each dimension ?
Maybe I don't understand something but when we choose to split X,Y,Z on the C2C algorithm we should have sign for each dimension, is'n it ? But it is not the case these values are always positive...
Let's take an example, if we have a point A(3,6,6) in a first point cloud (reference) and B(3,6,2) the closest point of A in a second point cloud (to compare). Then the absolute distance is D=sqrt(0+0+16)=4; and the coordinates of the vector AB are (0,0,-4) so on the Z axis there is a distance of -4 (i.e 4 units down).
Is there a way to have this information : the real (and not absolute) coordinates of the vectors between the closest points of two point cloud ?
I am working on tree growth. Using this information (i.e vector coordinates) for two point cloud of a tree at several years interval I can figure if the tree grows (or at least displace) more horizontally than vertically or the contrary...

Thanks a lot
best regards
Olivier

[daniel]

The X, Y and Z distance are signed when computed with the 'split X, Y and Z components' option.

I just checked with the latest version (2.6.3.beta) and it works fine. Are you sure you display the right scalar field?

Otherwise, don't hesitate to send me your clouds (or a small subset).

[olivier.martin]

Hello Daniel,

I just installed the latest version and it works well.
Thank you very much.
Olivier

PS: note that it doesn't work for the version 2.6.1 (i.e X, Y and Z are always positive).

[Saratchandra]

Hello everyone. I am a new user of cloud compare and I am trying to compare cloud to cloud distances. I have two point clouds overlapped on each other with a certain common area. The values are as shown in the figure and I would like to know the value of c2c distances instead of a bar. Can you please let me know how I can do that ? Thanks!

[daniel]

If you want to see particular distance values associated to particular points, you can use the 'shift + left click' shortcut to spawn a label on a point (or use the 'Point picking' tool: http://www.cloudcompare.org/doc/wiki/in ... nt_Picking).

You can change the colors/parameters of the currently displayed scalar field (see http://www.cloudcompare.org/doc/wiki/in ... properties and http://www.cloudcompare.org/doc/wiki/in ... es_Manager) to get a better visualization of what's happening in the part of interest. You can also extract the points below a given distance (or equivalently set a maximum distance when you compute the distances).

Otherwise you can also export all the values (with the points) if you export save the cloud as an ASCII file.