compare 2 point clouds

Feel free to ask any question here
yangyuanyuan
Posts: 9
Joined: Mon Oct 15, 2012 8:35 am

compare 2 point clouds

Post by yangyuanyuan »

Hi
If I want to compare 2 point Clouds,How to Quantitative analysis the deformation values. and how can I get the direction information(X,Y,Z ).
thank you very much!

yaoyanli
daniel
Site Admin
Posts: 7707
Joined: Wed Oct 13, 2010 7:34 am
Location: Grenoble, France
Contact:

Re: compare 2 point clouds

Post by daniel »

You should check the "split X,Y and Z components" in the 'General parameters' section of the Cloud-to-cloud distance computation dialog. You'll get a scalar field for each dimension.

And for further analysis you can export all those fields as an ASCII file (just select the compared cloud and use 'File > Save'). You can easily open this kind of file in Matlab, Excel, etc.
Daniel, CloudCompare admin
Sarhat
Posts: 5
Joined: Thu May 30, 2013 11:36 am

Re: compare 2 point clouds

Post by Sarhat »

I don't wanted to open a new thread, as there are lots of threads and my question is on comparison of point clouds or clod to mesh or vice versa.
I am a PhD student and I use different tools and software, my subject is on digital images for road surface monitoring.
I have found CC a very useful tool and many thanks for Daniel and this open source sponsor.
My question is I want to know the theory behind cloud to cloud distance computation.
I know the idea is simple but I could not find much details about it on this forum and on documentation.
I am not sure how CC compute these distances, is it closest distance between two points on the two clouds or the vertical distance between the two surfaces.
I highly appreciate if some one guide me to a theory or a link behind this theory.
Thanks
daniel
Site Admin
Posts: 7707
Joined: Wed Oct 13, 2010 7:34 am
Location: Grenoble, France
Contact:

Re: compare 2 point clouds

Post by daniel »

You must have missed the wiki:
http://www.cloudcompare.org/doc/wiki/in ... omputation

And for more theory about cloud-to-cloud distance:
http://www.danielgm.net/phd/isprs_laser ... 05_dgm.pdf
And for cloud-to-mesh distance:
http://vcg.isti.cnr.it/activities/surfa ... metro.html
Daniel, CloudCompare admin
freemountain
Posts: 11
Joined: Mon Jan 13, 2014 10:05 am

Re: compare 2 point clouds

Post by freemountain »

Is there a way to differentiate between negative and positive change of a surface - visually? Depending on the chosen color scheme there is bright area where change is biggest, independently from the direction. Can you help with that? I want negative change in one and accumulation in the other color.
daniel
Site Admin
Posts: 7707
Joined: Wed Oct 13, 2010 7:34 am
Location: Grenoble, France
Contact:

Re: compare 2 point clouds

Post by daniel »

If you compute distances relatively to a mesh, the distances are 'signed' by default (thanks to the triangle normals that indicates which side is outside or inside). Then you can edit the color ramp to fit your needs (see the "Edit > Scalar Fields > Color Scales manager" tool: you can create an "absolute" scale with different colors on each side of 0).

Otherwise, cloud/cloud distances are only positive (as there's no information on the side by default). A plugin should soon be ready (M3C2 - see http://nicolas.brodu.net/fr/recherche/m3c2/index.html) that will allow for signed distances between clouds but it will only be ready in a few weeks.
Daniel, CloudCompare admin
freemountain
Posts: 11
Joined: Mon Jan 13, 2014 10:05 am

Re: compare 2 point clouds

Post by freemountain »

Thanks again for your always quick responses, do you live at your desk? ;)
I now wonder if you can set the mean plane to pass through the min. point (from max min tool) so that you get the absolute z values in reference to a artificial plane which should be z=0 but still parallel to the pointcloud or the mesh
daniel
Site Admin
Posts: 7707
Joined: Wed Oct 13, 2010 7:34 am
Location: Grenoble, France
Contact:

Re: compare 2 point clouds

Post by daniel »

Oh I forgot: if you have a signed scalar field, you can also simply check the 'symmetrical color scale' option in the 'Parameters' tab o the SF display parameters (bottom section of the cloud 'Properties'). It will force the current (relative) color scale to be centered on 0.
Daniel, CloudCompare admin
daniel
Site Admin
Posts: 7707
Joined: Wed Oct 13, 2010 7:34 am
Location: Grenoble, France
Contact:

Re: compare 2 point clouds

Post by daniel »

No I simply hold my 'desk' in my pockets ;)

And I'm not sure to understand what plane you are referring to? (the min/max tool is only meant to segment points based on their associated scalar value). Can you be more specific? (maybe with a snapshot of your cloud?)
Daniel, CloudCompare admin
freemountain
Posts: 11
Joined: Mon Jan 13, 2014 10:05 am

Re: compare 2 point clouds

Post by freemountain »

I am just trying to find the difference between the mean plane and rough plane... that is above and below. Easiest would be to let that plane run through the min. value so that all other differences are above that plane. Important here is, though, that the plane needs to be parallel to that 'rough' plane that I have. I would also be happy with the result of just making a plane out of that mean plane so that it appears in the left window for further processing like a second pointcloud.
Merci beaucoup encore une fois!

Update: I think I figured it out myself by using the dropdown from the original pointcloud. Its still calculating the difference, though. Octree at 9;)
Post Reply