Automate alignment of two point clouds

[ertzu]

Hello everyone,

I manually align two point clouds and then compare them. I want to use CloudCompy to automate this process, is the alignment of two point clouds even possible without a GUI?

Thanks!

[daniel]

That's very probably possible. But it's better to ask questions directly on the Github project page: https://github.com/CloudCompare/CloudComPy/issues (I'm not sure Paul reads the forum messages, as CloudCompy is somehow a separate project)

[osnoopyo]

I too require this, I've had varied success.

can anyone share a basic workflow for 2x point clouds to align, scale, compare differences? In my case, my object will wear over time, I need to compare the original scan with the subsequent scan and identify the areas of wear.

The files are e57.

Any tips appreciated.

[daniel]

Same answer then ;)

[osnoopyo]

sorry, I misunderstood the particulars around the OP.
I want to do this within Cloudcompare, not cloudcompy.
Perhaps I should just open a new thread?

Basically, there are some reference points on the objects that do not wear but much of the objects does wear. I need the algorithm to automatically align everything and display the deviance histo and images.
I cannot yet get it to work reliably in the GUI but long-term need is to make it work in headless mode on a server and export the results to DB.
Is there are general recommendation as to steps in workflow?
EG: Load both point clouds, centre by bounding boxes, scale, align with ICP........ or some such sequence?

Thanks for your support.

[daniel]

Well, if the two clouds have roughly the same shape and extents, then yes, your workflow is good. I would just insist on the 'overlap' parameter for the ICP algorithm. This should not be too high otherwise some 'worn' points will be used in the matching process. And it should also take into account non-overlapping parts of the compared cloud of course.

Also, not sure which scaling you apply before ICP? Or did you mean the 'adjust scale' option of ICP? (this might not be necessary, or even counterproductive, depending on how the clouds are generated).