Differences in height for 2 point clouds from laser scanner

[Adam]

Hello,

I have a question. If the height (z 1value) of mesh 1 and the height (z 2 value) of mesh 2 are the same does mean there no differences in heights between the two meshes and M3C2 distance is the differences in height?

Example:
The averages/mean height of the first scan is z1= -17.159 mm and the second scan which is z2= -17.154 mm and the difference in height average/mean (M3C2 distance) is approximately 0.003 mm. we could say here the differences in height between the 2 scans is 0.003 mm which is small differences but it can be significant because of the large data, is this right? what if the M3C2 distance looks like the figure below does the image look right to you?

Thanks

[daniel]

Not sure what you mean by "it can be significant because of the large data".

But yes, if you select purely vertical normals, the M3C2 distance will be the difference in height. And this difference is 'averaged' in a a way, especially if you take a large 'scale' to compute it.

Regarding whether the image looks good or not, I would need a color scale attached to it... Bu I guess blue and brown are very close?

[Adam]

Hi, thanks for your reply. I meant because it's a dense point clouds and they have so many points.so any small difference is significant. But sometimes there are no differences in height as I get sometimes average height for both scans was -22.06 mm but the maps will show there is a difference in the topographic features. I don't know what to say in this case.
Yes, I have used vertical in the normal option.
So the average of the M3C2 will the difference in height which is around 0.003 mm? the thing which confuses me is the M3C2 contour maps showing a big difference but the number is small like 0.003 mm.

Yes, blue and brown are close. Blue is -0.1 and brown is 0 (I have attached the clour scale of the M3C2)

here is scan 1 and scan 2 and the M3C2 distance is the one that I have attached earlier

Thanks

[daniel]

So the average of the M3C2 will the difference in height which is around 0.003 mm?

Sorry, I didn't understand your question.

the thing which confuses me is the M3C2 contour maps showing a big difference but the number is small like 0.003 mm.

What do you mean by "big difference" ? And if your cloud is dense, pay attention that some points may hide other points...

[Adam]

I mean if I take the averages of M3C2 distance that generated from M3C2 computation like the value below is this the difference in heights?
M3C2_distance
-0.025911
0.036401
0.004997
-0.020805
0.043254
0.04156
0.042982
0.003481
-0.055137
-0.053322
-0.063224
-0.040197
0.009203
-0.016254
0.024587
0.033783
0.041482
0.04428
---------
average= 0.002842222 is this value is the difference (M3C2)
The values below are XYZ - Z values only which represent the heights
Z 1 (for Mesh 1, first scan) Z 2 (for mesh 2, second scan)
-17.6369 -17.5604
-17.6922 -17.5587
-17.7385 -17.5242
-17.5968 -17.5226
-17.6249 -17.6099
-17.6557 -17.6265
-17.6679 -17.6641
-17.6666 -17.6635
-17.5824 -17.6613
-17.5818 -17.6319
---------- -----------
Averages= -17.6477 mm Average= -17.6066 mm

First scan average height is -17.6477 mm
Second scan average height is -17.6066 mm
The difference in height which the M3C2 is 0.002842222 mm? right?

I meant by ''big differences'' is the features/pattern topography of the surfaces of both scans looks different (they are the same surface but they were scanned in two different periods like after 2 months, see the two contour maps of the surface) because the surface has changed over time I thought the height differences (M3C2 distance) will be higher than 0.002842222 mm

I hope my explanation is clear.
Thanks

[daniel]

I don't understand what you are trying to do or obtain with this 'average' values.

The difference of the height averages is not directly linked to the average of the M3C2 distances... The M3C2 distance is more complex (it has in itself a kind of local averaging effect to be more robust to noise, but it's another story).

Instead of averages, you should look at the histogram of the M3C2 distances for instance. It would be more meaningful...

[Adam]

Hi, thanks for your reply. I kind of understand now that I should not take the averages of the height of mesh 1 and 2, and compare them to M3C2 distances because of noises. So I will remove the averages of z values, but I want to know the difference in heights between the 2 meshes in numbers. For example, I want to be able to say 0.005 mm is the differences in heights which is the M3C2 distances. By looking at the histogram of the M3C2 below, Can I say the mean is the average height (M3C2 distance) is 0.002702 mm if not how can I be able to say this the difference in height with numbers?

Many thanks

[daniel]

Using a single number doesn't make sense when you have many points and distance values.

The best you can say is exactly what appears at the top of the histogram: 0.003 average and 0.045 standard deviation (unless you have an excellent accuracy, I wouldn't use more digits - with 3 digits you''l get microns already!).

The average alone means nothing if your standard deviation is of +/-1 mm. It only indicates a potential shift between the two subset (but once again, it's less than 3 microns!). The standard deviation gives a better idea of the actual differences.

[Adam]

Thanks

But how to do the same for each mesh? like histogram to be able to say this is the mean/std as it shows in the histogram for M3C2?
what are the important data to include from the M3C2 distance computation? I have used STD_cloud 1& 2 for the local roughness only
Npoints_cloud1
Npoints_cloud2
STD_cloud1
STD_cloud2
significant_change
distance_uncertainty
M3C2_distance