compare clouds for deformation-analysis

[daniel]

Oh, maybe you just miss the EDL shader ;) (see the 'Display > Shaders & filters' menu).

[kayillerwhale]

Hey Daniel,
you were right! it was just the missing shader :)

though i would be really thankful if you could have a look at the volume-calculation for the 2nd standpoint. after croppping the clouds to the same extends i get some really weird calculations. i know i'm really annoying but i need your help :(

6-7h:

Volume: -11.420
Surface: 99.919
----------------------
Added volume: (+)0.002
Removed volume: (-)11.421
----------------------
Matching cells: 93.0%
Non-matching cells:
ground = 3.5%
ceil = 3.5%
Average neighbors per cell: 7.5 / 8.0



6-8h:

Volume: -3.428
Surface: 99.977
----------------------
Added volume: (+)0.005
Removed volume: (-)3.433
----------------------
Matching cells: 93.3%
Non-matching cells:
ground = 3.4%
ceil = 3.2%
Average neighbors per cell: 7.5 / 8.0



6-9h:

Volume: -11.774
Surface: 99.557
----------------------
Added volume: (+)0.001
Removed volume: (-)11.775
----------------------
Matching cells: 92.8%
Non-matching cells:
ground = 3.8%
ceil = 3.4%
Average neighbors per cell: 7.5 / 8.0



6-10h:

Volume: 1.777
Surface: 99.814
----------------------
Added volume: (+)1.796
Removed volume: (-)0.019
----------------------
Matching cells: 93.2%
Non-matching cells:
ground = 3.5%
ceil = 3.2%
Average neighbors per cell: 7.5 / 8.0



6-11h:

Volume: -5.679
Surface: 102.869
----------------------
Added volume: (+)0.003
Removed volume: (-)5.682
----------------------
Matching cells: 95.5%
Non-matching cells:
ground = 0.7%
ceil = 3.8%
Average neighbors per cell: 7.7 / 8.0


i really don't know what's happening there :S the volume-calculation jumps from -11 to +1 to -3.. also the surface at 11h is pretty strange. it's 3qm bigger than all the others. i haven't continued for 12 and 13h so far.

Edit:

ok i just had a look and compared the clouds graphically. they are shifted to each other. but how could this happen? here you can clearly see the shift between 6 and 7h. the scanner + tripod didn't move. otherwise i would have recognized that the level moved and the scanner isn't placed horizontal anymore. but that wasn't the case. any other suggestions? does it make sense to register the shifted clouds first and then do the m3c2 + volume-calculation?

http://www.directupload.net/file/d/4450 ... tk_jpg.htm


Here you can see that even after registrating two scans (6-7) it isn't really done precisely.

http://www.directupload.net/file/d/4450 ... vf_jpg.htm
http://www.directupload.net/file/d/4450 ... 56_jpg.htm





Edit_2:

6h-12h

Volume: -10.463
Surface: 102.711
----------------------
Added volume: (+)0.002
Removed volume: (-)10.464
----------------------
Matching cells: 95.2%
Non-matching cells:
ground = 0.9%
ceil = 3.9%
Average neighbors per cell: 7.7 / 8.0


6h-13h

Volume: -10.121
Surface: 97.965
----------------------
Added volume: (+)0.002
Removed volume: (-)10.123
----------------------
Matching cells: 91.3%
Non-matching cells:
ground = 5.3%
ceil = 3.4%
Average neighbors per cell: 7.5 / 8.0

[kayillerwhale]

Since the volume-calculation at 7h (volume: -11.420) and 9h (volume: -11.774) was quite similar, i did a comparison between these two clouds:

Volume: -0.407
Surface: 99.832
----------------------
Added volume: (+)0.158
Removed volume: (-)0.565
----------------------
Matching cells: 93.2%
Non-matching cells:
ground = 3.6%
ceil = 3.2%
Average neighbors per cell: 7.5 / 8.0

as the calculated volume (-0.407) is likely to the results for the first standpoint, it's another sign that the scanner didn't move (imo).

[daniel]

Pay attention that the average number of (filled) neighbors is only 7.5 (ideally it should be 8 or almost 8). You can increase the grid step to reduce this effect (otherwise it may hinder the results).

[kayillerwhale]

Hi Daniel,
i will try it in an hour but it still doesn't justify the big shifts between the scans. at least for the m3c2-distances there will be huge differences.. i don't know what happened there.

[daniel]

Yes but I still don't have access to your data ;)

At least with more reliable volumes we will have a better basis to continue.

[kayillerwhale]

Ah ok ;)

So here are the results with an increased grid-size toi 0.04:

6-7h:

Volume: -11.677
Surface: 100.179
----------------------
Added volume: (+)0.001
Removed volume: (-)11.678
----------------------
Matching cells: 99.2%
Non-matching cells:
ground = 0.4%
ceil = 0.4%
Average neighbors per cell: 7.9 / 8.0



6-8h:

Volume: -3.504
Surface: 100.202
----------------------
Added volume: (+)0.002
Removed volume: (-)3.506
----------------------
Matching cells: 99.4%
Non-matching cells:
ground = 0.3%
ceil = 0.3%
Average neighbors per cell: 7.9 / 8.0



6-9h:

Volume: -12.082
Surface: 100.078
----------------------
Added volume: (+)0.001
Removed volume: (-)12.083
----------------------
Matching cells: 99.2%
Non-matching cells:
ground = 0.5%
ceil = 0.4%
Average neighbors per cell: 7.9 / 8.0



6-10h:

Volume: 1.823
Surface: 100.158
----------------------
Added volume: (+)1.830
Removed volume: (-)0.008
----------------------
Matching cells: 99.3%
Non-matching cells:
ground = 0.4%
ceil = 0.3%
Average neighbors per cell: 7.9 / 8.0



6-11h:

Volume: -5.632
Surface: 100.408
----------------------
Added volume: (+)0.002
Removed volume: (-)5.634
----------------------
Matching cells: 99.5%
Non-matching cells:
ground = 0.2%
ceil = 0.4%
Average neighbors per cell: 7.9 / 8.0



6-12h:

Volume: -10.390
Surface: 100.368
----------------------
Added volume: (+)0.001
Removed volume: (-)10.391
----------------------
Matching cells: 99.4%
Non-matching cells:
ground = 0.2%
ceil = 0.5%
Average neighbors per cell: 7.9 / 8.0


6-13h:

Volume: -10.465
Surface: 99.110
----------------------
Added volume: (+)0.001
Removed volume: (-)10.466
----------------------
Matching cells: 98.2%
Non-matching cells:
ground = 1.4%
ceil = 0.4%
Average neighbors per cell: 7.9 / 8.0

[kayillerwhale]

the surface values are kind of steady (differences caused by the empty cells again i guess) but as you can see the volume-calculation differs a lot.


edit: i just recognized that the 13h-cloud has a defect. there are some missing points because of a ship which covered a part of the wall. that's why the surface is smaller there. so i leave this cloud out.

[kayillerwhale]

i just had a look on the locations of the compared clouds (7-13) relating to the reference cloud (6h):


7h:
behind + left shifted

8h:
a little bit behind + minimal shifted to the left

9h:
nearly identical to 7h

10h:
nearly identical to 6h tiny bit in front and a bit to the right

11h:
a bit behind and shifted to the left (kind of similar to 8h)

12h:
behind and left shifted (nearly identical to 7h)

13h:
behind and left shifted (nearly identical to 7h)


but as i already mentioned above: i have absolutely no clue why the clouds are shifted as the scanner-tripod didn't move. it was fixed and in level all the time.

[kayillerwhale]

here are the results of the volume-calcuation AFTER registrating the clouds to the 6h-cloud.

6-7:

Volume: -0.005
Surface: 90.133
----------------------
Added volume: (+)0.176
Removed volume: (-)0.181
----------------------
Matching cells: 99.5%
Non-matching cells:
ground = 0.2%
ceil = 0.2%
Average neighbors per cell: 7.9 / 8.0



6-8:

Volume: -0.011
Surface: 90.093
----------------------
Added volume: (+)0.169
Removed volume: (-)0.180
----------------------
Matching cells: 99.5%
Non-matching cells:
ground = 0.3%
ceil = 0.2%
Average neighbors per cell: 7.9 / 8.0



6-9:

Volume: 0.004
Surface: 90.086
----------------------
Added volume: (+)0.180
Removed volume: (-)0.176
----------------------
Matching cells: 99.5%
Non-matching cells:
ground = 0.3%
ceil = 0.2%
Average neighbors per cell: 7.9 / 8.0



6-10:

Volume: 0.004
Surface: 90.091
----------------------
Added volume: (+)0.180
Removed volume: (-)0.176
----------------------
Matching cells: 99.5%
Non-matching cells:
ground = 0.3%
ceil = 0.2%
Average neighbors per cell: 7.9 / 8.0



6-11:

Volume: 0.003
Surface: 90.280
----------------------
Added volume: (+)0.156
Removed volume: (-)0.152
----------------------
Matching cells: 99.7%
Non-matching cells:
ground = 0.1%
ceil = 0.3%
Average neighbors per cell: 7.9 / 8.0



6-12:

Volume: 0.004
Surface: 90.277
----------------------
Added volume: (+)0.164
Removed volume: (-)0.161
----------------------
Matching cells: 99.7%
Non-matching cells:
ground = 0.1%
ceil = 0.3%
Average neighbors per cell: 7.9 / 8.0


6-13:

Volume: 0.000
Surface: 89.310
----------------------
Added volume: (+)0.193
Removed volume: (-)0.193
----------------------
Matching cells: 98.6%
Non-matching cells:
ground = 1.2%
ceil = 0.2%
Average neighbors per cell: 7.9 / 8.0