Field verification of Point Clouds

To chat about anything else.
Post Reply
kolsen
I have made <0 posts
I have made <0 posts
Posts: 3
Joined: Mon Aug 23, 2021 2:47 pm
2
Full Name: Kevin M Olsen
Company Details: Construction
Company Position Title: Technical Engineer Foreman
Country: United States
Linkedin Profile: Yes

Field verification of Point Clouds

Post by kolsen »

I currently use a Faro 350 S with Faro Scene 2021.4.0. Just recently scanned a 25,000 sq. ft. floor plate for the purposes of floor leveling. It was only 28 scans with 13 black and white targets tied to survey control and targetless registration utilizing planes and black and white targets. The entire perimeter was glass and that may be part of the problem. I couldn't get the overall point cloud to hit all of my black and white targets so I registered separate clusters to 3-5 targets each with an average dist. error of 4mm or less. Once the project point cloud is created and a 5x5 grid of points is extracted, we stake the points in the field to verify the scan data. I am typically seeing variations of +/- 0-12mm. A few questions/comments below.

1. Is anyone else doing this and what kind of results are you getting?
2. Any recommendations on registration techniques that may improve accuracy. (I typically do a top view based, followed by a cloud to cloud whilst calculating the target statistics. I start with average subsample of 63.09mm(.207 ft.) and search distance of 3.048m (10 ft.) and then decrease incrementally to see if registration results improve. If it gets worse, I revert to the previous setting and use a clipping box to check overlap both vertically and horizontally.

I have been on this forum before to research topics, but have never posted before so I thank you in advance for any comments/suggestions that may help in this endeavor. I look forward to connecting with other specialists in the industry to help improve my skills.

Kevin Olsen
Technical Engineer Foreman (Surveyor)
User avatar
jcoco3
Global Moderator
Global Moderator
Posts: 1724
Joined: Sun Mar 04, 2012 5:43 pm
12
Full Name: Jonathan Coco
Company Details: Consultant
Company Position Title: Owner
Country: USA
Linkedin Profile: No
Has thanked: 70 times
Been thanked: 157 times

Re: Field verification of Point Clouds

Post by jcoco3 »

Hi Kevin, welcome to the forum :D

The entire perimeter was glass and that may be part of the problem.
Oh yes that could be a problem on many fronts. For certain, glass is very large problem for use in C2C registration, and is next to impossible to properly utilize as registration geometry without extensive coating. Probably best to delete the glass and reflections from all your scans before updating your registration with whatever geometry may be left on the floor and ceiling.

Also, you did not mention as to whether or not your B&W checkerboard targets were mounted on this glass? If you used regular printer paper mounted directly to glass you could have some additional reflectivity issues due to the glass substrate or background for lack of a better word. Printed paper targets have partial opacity to the IR light that most scanners use, and the substrate that the paper target is mounted upon can considerable contribute to target reflectivity and acquisition. I have run into numerous problems with simple checkerboards mounted directly to glass or foam, and almost always try to mount the to fully opaque surfaces but even then the results can vary. It is probably better to use a manufactured checkerboard that better resist penetration of the laser or pre-mount your printed checkerboards on fully opaque backgrounds if you need to mount them on poor backgrounds like glass or foam. It is not always a problem, but a good thing to test with your targets in the office and to keep in mind when mounting them on site.
kolsen
I have made <0 posts
I have made <0 posts
Posts: 3
Joined: Mon Aug 23, 2021 2:47 pm
2
Full Name: Kevin M Olsen
Company Details: Construction
Company Position Title: Technical Engineer Foreman
Country: United States
Linkedin Profile: Yes

Re: Field verification of Point Clouds

Post by kolsen »

Thank you for that. Yes, they were mounted on the glass, but we do purchase our targets from a manufacturer and they are very opaque. If I eliminated all the glass except for the panes that contain the black and whites that may increase the accuracy. I will have to try that and see how it goes. Thanks for the advice!
User avatar
jcoco3
Global Moderator
Global Moderator
Posts: 1724
Joined: Sun Mar 04, 2012 5:43 pm
12
Full Name: Jonathan Coco
Company Details: Consultant
Company Position Title: Owner
Country: USA
Linkedin Profile: No
Has thanked: 70 times
Been thanked: 157 times

Re: Field verification of Point Clouds

Post by jcoco3 »

Good to hear that you have quality targets to work with.
You can also cut out those remaining panes that are around each target leaving just the targets which may also help.
fobos8
V.I.P Member
V.I.P Member
Posts: 246
Joined: Sat Jul 07, 2018 9:19 pm
5
Full Name: Andrew
Company Details: NDC Surveys
Company Position Title: Surveyor
Country: Uk
Linkedin Profile: No
Has thanked: 56 times
Been thanked: 22 times

Re: Field verification of Point Clouds

Post by fobos8 »

kolsen wrote: Tue Nov 30, 2021 5:09 pm I currently use a Faro 350 S with Faro Scene 2021.4.0. Just recently scanned a 25,000 sq. ft. floor plate for the purposes of floor leveling. It was only 28 scans with 13 black and white targets tied to survey control and targetless registration utilizing planes and black and white targets. The entire perimeter was glass and that may be part of the problem. I couldn't get the overall point cloud to hit all of my black and white targets so I registered separate clusters to 3-5 targets each with an average dist. error of 4mm or less. Once the project point cloud is created and a 5x5 grid of points is extracted, we stake the points in the field to verify the scan data. I am typically seeing variations of +/- 0-12mm. A few questions/comments below.

1. Is anyone else doing this and what kind of results are you getting?
2. Any recommendations on registration techniques that may improve accuracy. (I typically do a top view based, followed by a cloud to cloud whilst calculating the target statistics. I start with average subsample of 63.09mm(.207 ft.) and search distance of 3.048m (10 ft.) and then decrease incrementally to see if registration results improve. If it gets worse, I revert to the previous setting and use a clipping box to check overlap both vertically and horizontally.

I have been on this forum before to research topics, but have never posted before so I thank you in advance for any comments/suggestions that may help in this endeavor. I look forward to connecting with other specialists in the industry to help improve my skills.

Kevin Olsen
Technical Engineer Foreman (Surveyor)
Hi Kevin
If you have plenty of targets on the job, why don't you try a target base registration and then see how will the registered point cloud fits to survey control? You don't say anything about the nature of the area, is there a ceiling, and lots of features that would help a cloud to cloud registration.
Regards, Andrew
kolsen
I have made <0 posts
I have made <0 posts
Posts: 3
Joined: Mon Aug 23, 2021 2:47 pm
2
Full Name: Kevin M Olsen
Company Details: Construction
Company Position Title: Technical Engineer Foreman
Country: United States
Linkedin Profile: Yes

Re: Field verification of Point Clouds

Post by kolsen »

Thank you for the reply. The space is the interior of a high rise building with a central drywalled core and elevator lobbies. The entire perimeter is a glass curtainwall system. Using the black and white targets on exterior windows and on the central core for survey control and using planes from the core and other common surfaces to use targeted registration. Once that is run, I will "tighten up" the registration with the c2c while calculating target control as well. I am trying to eliminate all the exterior curtainwall points to see if this will work.
User avatar
TommyMaddox
V.I.P Member
V.I.P Member
Posts: 514
Joined: Mon Apr 25, 2016 7:15 pm
7
Full Name: Tommy R Maddox III
Company Details: ONSITE3D
Company Position Title: Director of Technology
Country: USA
Linkedin Profile: Yes
Location: Calgary, Alberta
Has thanked: 67 times
Been thanked: 130 times

Re: Field verification of Point Clouds

Post by TommyMaddox »

I've got a good bit of this work under my belt.

Lets chat when you get a chance, I can show you some pointers on the reg for analysis workflows.
Post Reply

Return to “General Chat”