Hello,
On a recent project I encountered an issue with the registration of the scans done for an elevator shaft. The scanner was set up at different levels to be able to get the surfaces of all the elevator shaft. The plan was to measure the difference between the as-built and the design. However, as seen in the attached images, the cloud shows 5 different layers of point clouds. The image is a slice of that area to double check the scans were aligned and that is what it shows. Also, a bump is shown with the scans in one of the images that is not as it was built. The registration of the cloud was to an average distance error of 4.7mm, but still if you measure between the layers of the point cloud there, it is 10-30mm. My questions are the following:
What is happening to the point cloud in this case?
What could have caused this issue?
How can it be prevented on the field?
How can I fix it to be able to have a point cloud that does not have multiple layers showing like that but a single surface scanned?
Thank you for your help,
Christian
Scan Point Clouds not Registered Correctly
-
- I have made <0 posts
- Posts: 2
- Joined: Thu Jan 17, 2019 3:10 pm
- 5
- Full Name: Christian Orozco
- Company Details: PCL - Construction
- Company Position Title: Field Coordinator
- Country: Canada
- Linkedin Profile: Yes
Scan Point Clouds not Registered Correctly
You do not have the required permissions to view the files attached to this post.
- Felix_the_Cat
- V.I.P Member
- Posts: 199
- Joined: Thu Jun 09, 2016 5:44 pm
- 7
- Full Name: Felix
- Company Details: Lockheed Martin
- Company Position Title: Senior Member Engineering Staff
- Country: USA
- Been thanked: 26 times
Re: Scan Point Clouds not Registered Correctly
Well, you haven't given much info to go on. How many scans? What kind of scanner? Is it brand new or a schlump a bump rental? What version of scene? How did you target? How did you register?
So I'm gonna take a stab and say you have a poor targeting approach. In fact, I'm going to guess that maybe you didn't target at all? It's all about the targets, baby.
An elevator shaft is the epitome of difficult scan job. Just symmetry in every direction. Any shape that can be created by an extrusion is a registration nightmare. So you have to target the crap out of it OR introduce dimensional control, which is even better. Do you have access to a total station or tracker?
I'll get out of the way and let the adults here give you a real answer. Good luck
Sent from my iPhone using Tapatalk
So I'm gonna take a stab and say you have a poor targeting approach. In fact, I'm going to guess that maybe you didn't target at all? It's all about the targets, baby.
An elevator shaft is the epitome of difficult scan job. Just symmetry in every direction. Any shape that can be created by an extrusion is a registration nightmare. So you have to target the crap out of it OR introduce dimensional control, which is even better. Do you have access to a total station or tracker?
I'll get out of the way and let the adults here give you a real answer. Good luck
Sent from my iPhone using Tapatalk
- landmeterbeuckx
- V.I.P Member
- Posts: 1616
- Joined: Tue May 01, 2012 5:19 pm
- 11
- Full Name: Lieven Beuckx
- Company Details: Studiebureau Beuckx
- Company Position Title: Owner
- Country: Belgium
- Linkedin Profile: Yes
- Has thanked: 183 times
- Been thanked: 548 times
Re: Scan Point Clouds not Registered Correctly
Like felix said control is the key here. These kind of things are extremely difficult to register with c2c as your tripod tends to have the same height all around the measurement and with similar scans all the time.
You have to place targets (checkerboards and spheres) at different heights from the floor level you're are on at the moment, close to a corner, farther away from a corner.
Survey control at the bottom and the top with in between some if possible. The best would be through a series of windows from the outside (if possible).
You have to place targets (checkerboards and spheres) at different heights from the floor level you're are on at the moment, close to a corner, farther away from a corner.
Survey control at the bottom and the top with in between some if possible. The best would be through a series of windows from the outside (if possible).
-
- I have made 10-20 posts
- Posts: 15
- Joined: Tue Apr 10, 2018 2:05 am
- 5
- Full Name: Andrew Beckett
- Company Details: Hellier Mcfarland
- Company Position Title: 3D Environments Manager
- Country: Australia
- Linkedin Profile: No
- Has thanked: 3 times
- Been thanked: 1 time
Re: Scan Point Clouds not Registered Correctly
I have noticed this issue also, I mentioned it to faro and they seemed surprised, They asked that I send them some sample data but I haven't had the time. I suggest registering fewer scans in each cluster, and checking scans visually as you register. Cloud to cloud should work in this environment, Targets should not be necessary. I don't know how the reports come up with good numbers when the clouds as obviously miss aligned.
-
- I have made 50-60 posts
- Posts: 59
- Joined: Sat Sep 01, 2018 1:19 pm
- 5
- Full Name: Jakub Kałaska
- Company Details: Geodezja3D
- Company Position Title: owner
- Country: Poland
- Linkedin Profile: Yes
- Location: Warsaw, Poland
- Has thanked: 17 times
- Been thanked: 7 times
- Contact:
Re: Scan Point Clouds not Registered Correctly
What c2c parameters did you used?
I would set up a control points with total station every 3th floor.
I would set up a control points with total station every 3th floor.
- Jaakko Tawast
- I have made <0 posts
- Posts: 8
- Joined: Wed Oct 07, 2015 12:51 pm
- 8
- Full Name: Jaakko Tawast
- Company Details: Mittasuora Oy
- Company Position Title: Service Manager
- Country: Finland
- Linkedin Profile: Yes
- Has thanked: 1 time
- Been thanked: 2 times
Re: Scan Point Clouds not Registered Correctly
You're using Faro scanner to scan the elevator shaft? If so you need a lot of control points as scanner doesn't have precise dual axis compensator.
I have done these using P40/targets/c2c/sometimes prisms, also in a high-rise building. It's easy to get some error but with control points and dual axis compensator it's fine.
I have done these using P40/targets/c2c/sometimes prisms, also in a high-rise building. It's easy to get some error but with control points and dual axis compensator it's fine.