Alignment of point clouds issue

[Cybid]

Hi,

Currently I am working on preparation of DTM from laser scanning (FARO) in order to prepare a simulaton of a car accident. I have the point cloud gained from 7 stations of scanner. Point cloud is processed in SCENE and exported as E57 format. On site I distributed several chessboards and surveyed them by total station. System coordinate of point cloud is established by one of the scanner station and other set of points is described by local coordinate system based on designated points (origin and direction of the axis).

Using "Align" tool in CC I uploaded the point cloud into the software and selected the center of the chessboards (6 points). Then I inserted the coordinates of equivalent points surveyed by total station. Finally I got the new position and orientation of point cloud but the point cloud is adjusted well in 2D but the differences appear in Z value (3D). It looks like the point cloud is upside down and instead of decreasing the Z values its conversely. Previously, I had a road above the embankment but after processing the road is below.

The same situation appears when I need to transform another point cloud to the same coordinate system established by the total station and when I display on both point clouds at the same time, the poinst are situated well in contiguity of reference points but the rest of point cloud is in wrong place. I mean when I transformed ground model to new coordinate system and I do the same for different point cloud exported from Agisoft as E57 (represent the trees, trails), the point clouds are not adjusted proper beside the reference points (the trees are in different position on the ground model then It is on site).

My question is: what can I do to make this aligment proper and which stage of my processing point cloud can couse this problem (opposite Z value and the point clouds are not fitted proper) ?

Regards,

Jakub

[daniel]

It looks like either a degenerate case where the points are in a very particular alignment (but it seems highly unlikely with 6 points and real coordinates). The other option is that either some points are inverted (i.e. wrong association of chessboards and total station coordinates) or that one coordinate system is 'indirect' (the cross product of X by Y is -Z instead of +Z).

If you can share with me your dataset along with the total station points I may have a more precise idea.

[Getarq]

Hi Jakub,
Why you don't use your control points directly on Faro Scene where I suppose you are aligning the clouds. You just have tu add a csv with the coordinate and orient the project! then you can export it in real coordinate!
One time something similar happened to me, well many times in deed, because of the problem described by Daniel!
The X and Y value are inverted.
Regards

Diego.

[Cybid]

Hi Jakub,
Why you don't use your control points directly on Faro Scene where I suppose you are aligning the clouds. You just have tu add a csv with the coordinate and orient the project! then you can export it in real coordinate!
One time something similar happened to me, well many times in deed, because of the problem described by Daniel!
The X and Y value are inverted.
Regards

Diego.

I was wondering to use SCENE for this purpose but I have no idea how to make it :/. I select one scan from all as reference and imported csv file with coordinates. Then I put points on chessboards on reference scan and choose option "correspondence" selecting the same point on scan and imported. After registration I've got a wrong results. So I think to do something invalidly.

Geteraq, do you have some PDF or link with video how to use coordinates from total station to get georeferenced final point cloud?

[Cybid]

The problem is solved. Just my app installed on a total station works invalidly and it coused the confusion. After surveying on site I can establish a new coordinate system based on choosing 4 points: 3 define X, Y axis and 4 - elevation. After using this option my coordinate system define opposite direction for Z axis because of the scalar. I multiplied Y and Z values by "-1" for total station data and all work fine.

Thank you for any support in this topic.