Controlling Accuracy of 3D laser Scans in a Large INTERIOR Space

Discuss FARO hardware here.
User avatar
smacl
Global Moderator
Global Moderator
Posts: 1002
Joined: Tue Jan 25, 2011 5:12 pm
11
Full Name: Shane MacLaughlin
Company Details: Atlas Computers Ltd
Company Position Title: Managing Director
Country: Ireland
Linkedin Profile: Yes
Location: Ireland
Has thanked: 416 times
Been thanked: 425 times
Contact:

Re: Controlling Accuracy of 3D laser Scans in a Large INTERIOR Space

Post by smacl »

Just to add my 2c to some great advice already given;
nyterydur wrote: Wed Jun 29, 2022 10:11 pmI guess from a surveying standpoint however, that's considered a reflectorless shot. This means that we are not shooting a prisim on a rod like normal and the signal is not being reflected back to the receiver to be "corrected". You are only relying on what your eyeball sees as far as aligning the crosshairs of the TS to be as exact as possible on the checkerboard. I'm being told that reflectorless shots on a TS are not as accurate, but maybe that's because our TS (which we do own) is quite dated. Not sure if newer technology has improved this.
Yes the shot is reflectorless, more importantly perhaps is the method is resection / free station. The most important thing here is getting good geometry and redundancy. Consider all the positions (stations) that you are likely to setup your scanner, mark at point on the ground at each of those positions, and place targets such that you have at least one visible target inside each quadrant of a circle surrounding that position, ideally not too close to the position. Then, if needed, add extra positions such that each target is visible from at least three station positions. Then survey in all the targets with a total station from each station position, take two or more rounds of measurement, take the results into a dedicated adjustment package such as STAR*NET (or even better SCC :D ) and you're done. Height of instrument is also very important here so make sure to get it right

Possibly worth noting that a good control survey for scanning can involve the same amount or more site work than the scanning itself. As others have said, if this is a one off situation, getting in a good surveyor will likely produce a better result at a lower cost than attempting it yourself for the first time. That said, it is a worthwhile skill to learn and should be well within the ability of those with a good amount of scanning experience. If you do use a surveyor, tell them not to use regular retro targets as they don't play well with some scanners.
CMounts
I have made 60-70 posts
I have made 60-70 posts
Posts: 63
Joined: Tue Aug 18, 2009 5:23 pm
13
Full Name: Chris Mounts
Company Details: PMC
Company Position Title: Scanning Director
Country: USA
Linkedin Profile: Yes
Been thanked: 5 times

Re: Controlling Accuracy of 3D laser Scans in a Large INTERIOR Space

Post by CMounts »

A lot of good advice here but as a non-surveyor who lived through the transition you are going through I may have some insight. First, and most importantly you did not give much information on what a "large" site is. If we are talking about sub-200k square feet you can do a lot with solid target-based scanning. In most cases, you can get acceptable accuracy in your registration. If you are talking 1m+ with a high accuracy requirement you need to do a lot more than even a traditional survey.

Several years ago my company was pushing the limits of what could be done with surveyless scanning. We seriously messed up some projects and damaged our reputation. Since then we've recovered and are now using some of the most advanced survey control practices around. The journey to get there took some time and here are the steps I'd recommend.
  • Find a surveyor you can build a long-term relationship with as a sub-contractor. Many who we worked with were not good. What I came to realize is that there seems to be a difference between exterior and interior surveying that I still don't fully understand.
  • Include this surveyor in your bid process. Don't ever assume that the survey will be the same just because your scanning is about the same.
  • Do many projects as a team with your surveyor. If the pricing and workload is working out, just stick with this relationship. Keep survey as a subcontracted service.
  • If you find that it is not working out, look to hire a surveyor with interior experience. Just understand that the amount of gear surveyors need can be a lot more than you'd expect. I've got cases of marking spray paint now. I don't even know why :lol:
To answer some of your questions directly,
Reflectorless - Checkboards are but they are usually taken as side shots from a primary control loop. The primary loop will be based on prisms.
Is survey common practice - On larger projects, yes. Without exception. To do otherwise is not responsible.
Can't set pins - This is not correct. You certainly can and surveyors do it all the time. This can be complicated if you have a space with strict dust requirements or similar but there are workarounds that deliver close to the same accuracy as setting points.

As I said, I've lived this transition in a very painful way and would be happy to answer any other questions you may have to help you avoid the pain.
maxwell566
I have made <0 posts
I have made <0 posts
Posts: 7
Joined: Sat Apr 23, 2022 12:45 am
Full Name: Sean Maxwell
Company Details: Hagen Surveys
Company Position Title: Owner
Country: Canada
Linkedin Profile: Yes

Re: Controlling Accuracy of 3D laser Scans in a Large INTERIOR Space

Post by maxwell566 »

ill put my 2c in here as i am a long time surveyor and new to scanning.

i have the newest Trimble S9 High precision model and its accuracy on reflecrtorless shots ( paper targets ) is 2mm by spec 1mm to prism. Through our testing i have proven that it is actually achievable providing the surveyor knows what they are doing and shoots the targets at a relatively good angle ( straight on as possible ) the beam diverges quite a bit as you go further away so shooting paper targets at a 90 angle is the best way. As mentioned above taking multiple readings in reverse instrument faces and running through starnet is going to give you the best result. But at the end of the day it depends what your trying to accomplish. One shot on a paper target from a total station is probably going to give you the results your looking for with 99% of projects without going totally overboard on survey.

every time you have to move the instrument it induces error into your survey. Realistically the most accurate survey will be done from one setup and putting in as many controls as possible for your scan from that one setup. Traversing through a building is less than ideal to put targets into individual rooms etc.

But now I am curious as to how everyone is using survey control in their workflows?

Scene,trw only allows me to georeference scans to survey control. there isnt a workflow to apply a transformation to the scans to re register them to the survey control ( as far as i can tell ) this would be ideal. Register the entire project cloud to cloud and then holding the survey control to adjust the entire scan as a whole to the survey control. i dont see that this exists anywhere.

how are you registering hundreds of scans to only a few survey controls if they aren't visible from all scans? newbie scanner here. i would assume you would register everything cloud to cloud and then select the targets and re register everything holding the targets?>
CMounts
I have made 60-70 posts
I have made 60-70 posts
Posts: 63
Joined: Tue Aug 18, 2009 5:23 pm
13
Full Name: Chris Mounts
Company Details: PMC
Company Position Title: Scanning Director
Country: USA
Linkedin Profile: Yes
Been thanked: 5 times

Re: Controlling Accuracy of 3D laser Scans in a Large INTERIOR Space

Post by CMounts »

Scene,trw only allows me to georeference scans to survey control. there isnt a workflow to apply a transformation to the scans to re register them to the survey control ( as far as i can tell ) this would be ideal. Register the entire project cloud to cloud and then holding the survey control to adjust the entire scan as a whole to the survey control. i dont see that this exists anywhere.
It's been a long while since I used Scene regularly but I believe what you are attempting to do can be done. You would register C2C and then lock the cluster before applying control.
how are you registering hundreds of scans to only a few survey controls if they aren't visible from all scans? newbie scanner here. i would assume you would register everything cloud to cloud and then select the targets and re register everything holding the targets?>
Assuming you are using a fixed scanner if a project is large enough to require control, it's large enough that you should be using targets. There are always exceptions but in general, a small C2C project can be manually aligned faster than the time needed to run control on-site.

For us, the amount of control scales based on the requirements of the project. In most cases, we would never have only a few control points. Dozens would be on the low end. In the most extreme cases, we install permanent benchmarks by the hundreds. The image shows the field notes for the installation of 576 washers that were all surveyed. They cover a bit over one million square feet.
control_network.jpg
You do not have the required permissions to view the files attached to this post.
Augusto 3D
V.I.P Member
V.I.P Member
Posts: 206
Joined: Sat Dec 06, 2014 5:26 am
7
Full Name: C-Augusto
Company Details: 3d scanning
Company Position Title: USA
Country: US
Has thanked: 44 times
Been thanked: 3 times

Re: Controlling Accuracy of 3D laser Scans in a Large INTERIOR Space

Post by Augusto 3D »

CMounts wrote: Fri Jul 01, 2022 12:52 pm
Scene,trw only allows me to georeference scans to survey control. there isnt a workflow to apply a transformation to the scans to re register them to the survey control ( as far as i can tell ) this would be ideal. Register the entire project cloud to cloud and then holding the survey control to adjust the entire scan as a whole to the survey control. i dont see that this exists anywhere.
It's been a long while since I used Scene regularly but I believe what you are attempting to do can be done. You would register C2C and then lock the cluster before applying control.
how are you registering hundreds of scans to only a few survey controls if they aren't visible from all scans? newbie scanner here. i would assume you would register everything cloud to cloud and then select the targets and re register everything holding the targets?>
Assuming you are using a fixed scanner if a project is large enough to require control, it's large enough that you should be using targets. There are always exceptions but in general, a small C2C project can be manually aligned faster than the time needed to run control on-site.

For us, the amount of control scales based on the requirements of the project. In most cases, we would never have only a few control points. Dozens would be on the low end. In the most extreme cases, we install permanent benchmarks by the hundreds. The image shows the field notes for the installation of 576 washers that were all surveyed. They cover a bit over one million square feet.
control_network.jpg
I dont want to know how long that took... but at the same time I do.
So are all those washers relative to a common origin / setup point? or are they relative to different areas within that building.

Some asked about transformations, in Scene if you UNLOCK your cluster, then it will transform to those distances. If you keep the cluster LOCKED then you are just moving your scan. I could be wrong.
VXGrid
V.I.P Member
V.I.P Member
Posts: 347
Joined: Fri Feb 24, 2017 10:47 am
5
Full Name: Martin Graner
Company Details: PointCab GmbH
Company Position Title: Research and Development
Country: Germany
Linkedin Profile: No
Has thanked: 112 times
Been thanked: 109 times
Contact:

Re: Controlling Accuracy of 3D laser Scans in a Large INTERIOR Space

Post by VXGrid »

maxwell566 wrote: Thu Jun 30, 2022 6:23 pm [...]

Scene,trw only allows me to georeference scans to survey control. there isnt a workflow to apply a transformation to the scans to re register them to the survey control ( as far as i can tell ) this would be ideal. Register the entire project cloud to cloud and then holding the survey control to adjust the entire scan as a whole to the survey control. i dont see that this exists anywhere.

how are you registering hundreds of scans to only a few survey controls if they aren't visible from all scans? newbie scanner here. i would assume you would register everything cloud to cloud and then select the targets and re register everything holding the targets?>
I guess it doesn't matter if you georeference to the local total station net you surveyed, or to a country coordinate system.
In our software we advice to go the road to include the survey points into the registration, since they have an accuracy as well and you want to have them in your least squares.

IMO doesn't make sense to register everything c2c, only to do a 6 parameter transformation onto your complete cloud, because then your constellation itself is less secured than if you do everything in one single step.
I mean: How do you align the measured total station point (checkerboard target) with your point cloud? If you do it as a separate 6 param transfo, then you will likely set only one single point, while if you register the scans together with it, every scan which "sees" the target will be appended to the total station survey.
User avatar
TommyMaddox
V.I.P Member
V.I.P Member
Posts: 458
Joined: Mon Apr 25, 2016 7:15 pm
6
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: 39 times
Been thanked: 83 times

Re: Controlling Accuracy of 3D laser Scans in a Large INTERIOR Space

Post by TommyMaddox »

CMounts wrote: Fri Jul 01, 2022 12:52 pm
Scene,trw only allows me to georeference scans to survey control. there isnt a workflow to apply a transformation to the scans to re register them to the survey control ( as far as i can tell ) this would be ideal. Register the entire project cloud to cloud and then holding the survey control to adjust the entire scan as a whole to the survey control. i dont see that this exists anywhere.
It's been a long while since I used Scene regularly but I believe what you are attempting to do can be done. You would register C2C and then lock the cluster before applying control.
how are you registering hundreds of scans to only a few survey controls if they aren't visible from all scans? newbie scanner here. i would assume you would register everything cloud to cloud and then select the targets and re register everything holding the targets?>
Assuming you are using a fixed scanner if a project is large enough to require control, it's large enough that you should be using targets. There are always exceptions but in general, a small C2C project can be manually aligned faster than the time needed to run control on-site.

For us, the amount of control scales based on the requirements of the project. In most cases, we would never have only a few control points. Dozens would be on the low end. In the most extreme cases, we install permanent benchmarks by the hundreds. The image shows the field notes for the installation of 576 washers that were all surveyed. They cover a bit over one million square feet.
control_network.jpg
I can vouch for their process. We've worked under their procedures on this type of project and it is exceptionally accurate and reliable.
User avatar
smacl
Global Moderator
Global Moderator
Posts: 1002
Joined: Tue Jan 25, 2011 5:12 pm
11
Full Name: Shane MacLaughlin
Company Details: Atlas Computers Ltd
Company Position Title: Managing Director
Country: Ireland
Linkedin Profile: Yes
Location: Ireland
Has thanked: 416 times
Been thanked: 425 times
Contact:

Re: Controlling Accuracy of 3D laser Scans in a Large INTERIOR Space

Post by smacl »

VXGrid wrote: Mon Jul 04, 2022 7:46 amIMO doesn't make sense to register everything c2c, only to do a 6 parameter transformation onto your complete cloud, because then your constellation itself is less secured than if you do everything in one single step.
I mean: How do you align the measured total station point (checkerboard target) with your point cloud? If you do it as a separate 6 param transfo, then you will likely set only one single point, while if you register the scans together with it, every scan which "sees" the target will be appended to the total station survey.
Agreed, a rigid transformation as a last step is liable to do more harm than good unless it includes enough control points to entirely envelop the point cloud. A transformation with weak control is liable to introduce error, particularly where you extrapolate outside of the bounding region of the control points.
nyterydur
I have made <0 posts
I have made <0 posts
Posts: 7
Joined: Fri May 19, 2017 2:13 pm
5
Full Name: Nate Vallade
Company Details: Matrix Technologies Inc
Company Position Title: Facilities Engineer 3
Country: USA
Linkedin Profile: No
Been thanked: 2 times

Re: Controlling Accuracy of 3D laser Scans in a Large INTERIOR Space

Post by nyterydur »

maxwell566 wrote: Thu Jun 30, 2022 6:23 pm Scene,trw only allows me to georeference scans to survey control. there isnt a workflow to apply a transformation to the scans to re register them to the survey control ( as far as i can tell ) this would be ideal. Register the entire project cloud to cloud and then holding the survey control to adjust the entire scan as a whole to the survey control. i dont see that this exists anywhere.>
maxwell566, the way we incorporate survey control into Scene that was ran as c2c is:
1. We have a scans cluster containing all the scans properly placed and locked using c2c method
2. We create a Control cluster at the highest level and drag the scans cluster into it
3. We create a References folder within the Control cluster and import surveyed points or manually enter in points / checkerboard coordinates
4. Mark those same targets (points or checkerboards) in the individual scans they can be seen in
5. Rename the marked point to match exactly what that corresponding point is called as entered in the references folder
6. Run target based registration on the Control cluster, and select Force correspondence by manual target name

What step 6 does is it forces the program to only look for matching correspondence names and forces those correspondences, its just a clean way of doing it knowing that your reference point is not trying to refer to the wrong marked point for any reason.
jedfrechette
V.I.P Member
V.I.P Member
Posts: 1136
Joined: Mon Jan 04, 2010 7:51 pm
12
Full Name: Jed Frechette
Company Details: Lidar Guys
Company Position Title: CEO and Lidar Supervisor
Country: USA
Linkedin Profile: Yes
Location: Albuquerque, NM
Has thanked: 43 times
Been thanked: 152 times
Contact:

Re: Controlling Accuracy of 3D laser Scans in a Large INTERIOR Space

Post by jedfrechette »

@nyterydur the problem with that methodology is that your overall accuracy is entirely dependent on how good the c2c registration is. As @smacl and others have pointed out, there are plenty of situations where the accuracy of an unconstrained c2c alignment will be very poor, e.g. tunnels.

Ideally what you would want to do is have the control points from your independent survey attached to the scans that they appear in with a tolerance so that you could run a c2c adjustment of the scans that is constrained by those control points. That allows you to take advantage of both the higher accuracy of your control points as well as the high degree of sampling provided by overlapping scans. Unfortunately, that's not really possible if you're relying on Scene for registration.
Jed
Post Reply

Return to “FARO”