compare clouds for deformation-analysis

[kayillerwhale]

btw: you'll definitely get a big "thank you"-spot in my bachelor-thesis :D dunno what i'd have done without you.

[kayillerwhale]

any ideas what could have caused the shifts?

[daniel]

hI looked at the scans and it's clearly visible that there are high shifts between the scans. As the displacement is huge (> 0.05), rigid (the wall scan is not distorted but simply shifted as a whole) and it happens at various epochs, I really think the scanner has moved.

From my point of view It's totally valid to register the scans as they are clearly shifted (and not distorted).

But considering that the clouds are wide and very flat, ICP (Fine Registration) can't work properly. You'll have to first use the 'Point pair based' registration tool (Align). Use the various corners and features that are clearly recognizable on both clouds. And use a lot of pairs, and spread them all over the clouds if possible (even though here they are mostly aligned in the central area sadly):

cc_align_wall.jpg (130.65 KiB) Viewed 2736 times

Then you can use ICP with a low overlap ratio (e.g. 60%) and a small RMS difference (e.g. 1.0e-6).

Of course it's a very manual process, but it's the best you can do without physical targets...

Here is what I get between 6h and 7h:

cc_m3c2.jpg (194.3 KiB) Viewed 2736 times

It's much better than the original positioning. And you still get the 'wave' effect even if it's quite small between these two clouds. I haven't tested the others.

P.S.: Not sure if I already mentioned this, but I find the clouds pattern a little odd (here is the cloud 11h when looked with a high zoom and from the top):

cc_odd_pattern.jpg (136.23 KiB) Viewed 2736 times

Have you applied any process before exporting the clouds from Cyclone?

[Dimitri]

Hi,

Which sensor and the setup did you use ?
Does the sensor has a compensator and was it activated ?
Did you use a metallic or a wood tripod (or even better, a concrete pilar) ?
What was the ground made of ?
For instance, a metal tripod will have significant thermal expansion (with respect to the deformation you're trying to measure) that will shift slightly the position and tilting of the sensor when changing from shadow to direct sunlight. Then the sensor itself could be slightly affected by the temperature.

Next time, you should absolutely place fixed targets around the scanner (even behind you) that you remeasure for each scan, and make sure that the scanner (and targets) are always in the shadow. As Daniel explained, scanning the building behind the wall is not enough to ensure a proper registration of the scans as you cannot precisely compute a registration (the solid angle corresponding to the building is actually extremely small for accurate registration).

Dimitri

[surda007]

Hi!

Can I apply the EDL shader on just severeal clouds of the open ones, not all of them?
Thanks!

[daniel]

Nope, this shader applies to the whole graphic view, not to particular entities.