Registration and error reporting

[armstrong_geology]

Hi Daniel,

More questions:

1. I have been able to register my cloud to my reference GPS survey very well using CC. I can easily get distance computations from CC that report everything I want to know. HOWEVER, I would like to do the analysis myself, so that I can make my own histograms etc. My solution to this is to do the analysis in ArcGIS and export the results as a table for manipulation in Excel. This is easy enough, however I've come to realize that either:
- I've registered the 2 clouds improperly, as their coordinates are not similar, or
- The registration in CC is not changing the coordinates, but only moving the cloud in its existing local coordinate system.
My GPS (reference) cloud is referenced to real world coordinates. Is there a function in cloud compare that will update the unregistered cloud coordinate system to the same one as the reference GPS survey? I had thought that that was what I was doing with point to point registration.

2. Also, I have interpolated the cloud outside CC using inverse distance weighting and wanted to compare it to the GPS using CC's cloud to mesh....what format or file type does the mesh need to be in to bring it to cloud compare? As a .ply it came in as a train wreck.

Thanks for any help!!
Lindsay

[daniel]

1. I guess your GP survey cloud was 'shifted' at loading time? If yes, and if you use the point pair based registration (latest version) then the 'shift' information should be automatically transferred to the aligned cloud. Otherwise (with ICP registration for instance - or in case things didn't work the expected way ;) you should check that the shift information on the aligned cloud is up-to-date (select the cloud and use "Edit > Edit Global Shift"). If not, you can simply copy the 3 values of the reference cloud.

(I'll update the ICP tool so that it copies the shift information automatically, just as the point pair based registration does).

2. Ply should be ok (I'm interested in your file!) but you can also use OBJ, STL or VTK formats.

[armstrong_geology]

Daniel,

Thanks for the speed of reply! Incredible!

So I have been able to get my cloud to have the proper coordinates (similar to the real world GPS), however they are not near one another in Arc. The GPS cloud was being multiplied by -1 in the x, but I simply removed them. Still, the GPS and cloud are not near each other!! Is it possible that although the cloud now has proper coordinates, Arc does not recognize it? I tried to reproject it in Arc, but without telling it the original coordinate system, it cannot do the math....makes sense....
Is there a way to output the point to point comparison as a .txt file, so I can see the data in excel? This would eliminate my issue....

Thanks again, and please, by all means let me know how I can send you anything you'd like to see! Also, I'd love to recognize you in my paper.
Is that ok? (Undergrad thesis)

Lindsay

[daniel]

I know very little about ArcGIS (I followed a training about 10 years ago, but at that time I don't believe that point clouds were handled at all - or at least I forgot it all ;).

It's very simple to export the distances (along with the points). Simply save the cloud as "ASCII" (and use semicolon as separator to get a true Excel "csv" file).

You can send me files to 'cloudcompare [at] danielgm.net' (or if its too big with YouSendIt, Google drive shared file, dropbox, etc.).

And about citation, see this post: viewtopic.php?t=146.

[armstrong_geology]

Hi Daniel,

I'm sorry to ask to be spoon fed, this is all new and magical! ha!

If I export the cloud, how will the distance between it and the GPS reference cloud be included? Wouldn't that then be xyz info (500000, 400000, 100), and small distances (0.0002)? I'm confused.

Also, you are right. Arc is not good at clouds. That's why you and CC are my oracle. :)

Lindsay

[daniel]

The distances will appear in a fourth colum (X Y Z dist). But indeed, the distance values might be very small so you'll have to use a very high precision.