Page 1 of 1

M3C2

Posted: Mon Sep 15, 2014 6:35 am
by jamestesoriero
Hi, Danial/Cloud Control Admin

Ive been playing around with your software and i was woundering if i could get some help? Ive been trying to compare two point clouds representing a wetland. One cloud shows the site at low tide and the other at high tide. I hoping to use cloud compare to show the difference in innudation between the two clouds, by using C2C tool and or the M3C2 plugin. So i was woundering firstly how does the 'guess params' function for the M3C2 plugin work in esimtating the scales and is it sufficent in estimating the approiate scale values e.t.c? And is there a way to remove the effect of vegetation, as its presences is affecting the output of my C2C anaylis.

Regards James Tesoriero

Re: M3C2

Posted: Mon Sep 15, 2014 7:58 am
by daniel
Hi,

Could you post a snapshot of the clouds so as to make things clearer maybe?

I'll also ask Dimitri (M3C2's creator) if he can answer your quetsions.

Re: M3C2

Posted: Mon Sep 15, 2014 8:28 am
by Dimitri
Hi James,

picture of your point clouds would help.

for M3C2:
in your case, the guess param will have a limited use because what you want to do is the vertical difference. in that case you tick in the "Normals" tab: vertical. Then you don't have to worry about the "normal scale".
As for the projection scale, the guess param could help, but in general it's not very complicate to set up. You must select it such that you have typically more than 5 to 10 points within a diameter of the projection scale. For instance, if your point density is aroun 1 pt/cm², then a projection scale of 10 cm would be ok. It all depends on the level of spatial averaging you aim for and the variability in topographic change you want to detect.
As for the depth, just take about twice the maximum distance between the 2 clouds that you expect.

As for removing the vegetation it all depends on how many points are actually hitting the ground below or between the plants. In densely vegetated area with single echo terrestrial lidar, there's no chance to hit the ground, hence no way to remove the vegetation. In sparsely vegetated areas, it's very easy to remove the vegetation to work only on the ground. Check the paper on CANUPO (Brodu and Lague, 2012) and grab classifiers and tutorials on that on my web page.

Dimitri

Re: M3C2

Posted: Mon Sep 15, 2014 1:34 pm
by jamestesoriero
Thanks for quick reply.
Thanks for the help so far. I’ve included some screenshots of the two point clouds, to make it clearer. The density of the two clouds is about 4cm/pixel. I guess I want to measure the smallest possible topographical change as possible, as the difference between the two water levels will be about 10-20cm. As you said I don’t think I can remove the vegetation in some areas, due to the density, or can I? If there is anything else you could add, it will be much appreciated.
Regards James Tesoriero

Re: M3C2

Posted: Wed Sep 17, 2014 12:54 pm
by daniel
As Dimitri suggested, the qM3C2 algorithm with 'vertical normal' mode is clearly the best way to go. And if you manage to fully remove the vegetation with the qCanupo plugin on one of the clouds (it may be simpler on the lowest tide cloud), it should avoid making false detections in the qM3C2 process? qCanupo is really easy to use.

On a side note, I intend to develop soon a tool to compute distances and volume difference along a single direction. This tool would be based on the same principle than the 'Rasterization' tool (Tools > Projection > Rasterize). Would you be interested in beta-testing it? Or can you share your data with me? (you can send me an email to cloudcompare [at] danielgm.net if you're interested).