Grid to Ground

Discuss all Construction & Architecture related issues here.
Post Reply
User avatar
bwiab
I have made 80-90 posts
I have made 80-90 posts
Posts: 88
Joined: Thu May 08, 2014 6:48 pm
Full Name: Emilio Ramirez
Company Details: Metro
Company Position Title: BIM Supervisor
Country: USA
Linkedin Profile: Yes

Grid to Ground

Post by bwiab » Wed Jul 02, 2014 6:40 pm

At what point does grid to ground become an issue for scanning? The campus that I'm scanning covers over a mile and a half. Should I be scaling all my scans?

bwx100
V.I.P Member
V.I.P Member
Posts: 138
Joined: Fri May 25, 2012 11:29 pm
Full Name: Bill Wallace
Company Details: Topcon Positioning Systems
Company Position Title: Mktg Strategy Mgr for Machine Control
Country: USA
Linkedin Profile: No

Re: Grid to Ground

Post by bwx100 » Wed Jul 02, 2014 7:49 pm

Great question. I would like to hear everyone's opinions too.

Also, if I was scaling would I leave each point cloud alone and just scale it's center point or should I scale the location of every point?

I am not so well versed in this area as to really speak to it at a precise level but I have sat in on the long debates of how those in the know think it should be done and I never feel there is any consensus.

User avatar
Phill
V.I.P Member
V.I.P Member
Posts: 586
Joined: Wed Feb 06, 2008 7:19 am
Full Name: Phillip Nixon
Company Details: SKM
Company Position Title: Surveyor
Country: Australia
Location: Sydney

Re: Grid to Ground

Post by Phill » Thu Jul 03, 2014 12:44 am

Emilio

To answer your question, that is for you to decide. What is the scale factor, and how much difference does it make over the distance you are measuring.

We often have to put terrestrial scan data from plane to MGA (grid). At it's maximium we can have a scale factor up to 0.9996. So here is an example of the potential difference (I do love math after all) and a few solutions.

If we are using a HDS7000 which has a usable range of about 50m the difference using a scale factor of 0.9996 would be 20mm.
If we do a line of scans over 1.5miles (2414m) the difference would be 0.966m (966mm)

At this point you need to decide if you want to run everything on grid or plane (ground) as it affects what you can do. I will mention at this point that cyclone has no ability to scale a cloud, raw or registered (that I know of). I am just going to assume you are using cyclone... if not you might be able to find another way.

Option 1. Scale the raw clouds
In the field make sure your total station is setup to measure your targets in grid. Once you have imported the scans export them as a ptx file and then get someone to write a tool to scale the ptx, then bring them back into cyclone for registering. You might be able to do this with the matirx at the top of the ptx (I've never tested that), or just do it on the x and y measurements.

Option 2. Do everything on plane and then scale at the end
Take your grid control coordinates and convert them to plane by picking a point and backscaling. Once you have made the registered point cloud, export the unified cloud to a pts file and then scale the pts. You will need use that backscale origin point to scale from, and keep all your total station work in plane.

Someone did ask Leica to be able to put a scale factor into Cyclone back in 2008 at the HDS conference, but they must have been very busy with other things, like automatic cloud to cloud functions and such. The programming is not that hard, if you choose option 2 you can use the excel file attached.

Bill, I see you mentioned that there has been some debate on this but I dont quite understand what the debate is about, if you scale the centerpoint (i.e. put the position on grid) all the measurements would still be out. It's like doing a topographical survey on grid, but forgetting to put the scale factor into your jigger. Especially as you are the suppliers of the scanners that have the 0-150m range this should be something that is important, not so much on the shorter range units. As far as I can see, either put the opition to scale the cloud on import, or make it available as a tool in the modelspace (i.e. Scale cloud... set as default cloud). Of course I am not on the programming team, but if I can find a workaround in a few minutes, perhaps your guys could have a quick look at it... then again, perhaps myself and Emilio... and about 10 other people I can think of at the moment, are the only ones that have ever had to deal with this problem. Maybe the scale factors in the US and Europe are not quite as large as ours.

Cheers

Phill
You do not have the required permissions to view the files attached to this post.

TMillington
I have made 50-60 posts
I have made 50-60 posts
Posts: 51
Joined: Wed Jun 12, 2013 10:59 am
Full Name: Tony Millington
Company Details: ISIS Rutherford Appleton Laboratory
Company Position Title: Survey section leader
Country: uk
Linkedin Profile: Yes

Re: Grid to Ground

Post by TMillington » Thu Jul 03, 2014 9:00 am

I’m sure this issue has been raised before, it’s a perennial one-for survey, not just scanning.

How ‘bothered’ you are about applying the scale factor will largely depend on what your client wants at the end.
If they want to measure between points anywhere within your data and get the distance they’ll get on the ground you could probably stick to a plane grid and ignore scale factors.

If your data has to be consistent with a national/state/ GNSS derived grid system, scale comes into play.

Ideally then, I’d go for Phill’s option 2 and compute/register everything on a plane grid before scaling or transforming every point to the final grid system.

As a compromise, the ‘easier’ option is to use control points with grid coordinates to register the scans. There will be a scale component in the registration errors for each scan but, depending on the distance covered by each scan, you may be able to live with the errors at the overlap of scans.
If the registration computation applied a scale correction as part of the adjustment, that would mitigate the scale problem, but I’m not aware of any of the standard packages that allow this.

Doubtless there's somebody out there who can suggest registration software that will cope with scale factors.

User avatar
danielgadowski
V.I.P Member
V.I.P Member
Posts: 451
Joined: Mon Jun 30, 2008 3:49 pm
Full Name: Daniel Gadowski
Company Details: London Underground
Company Position Title: Laser Scanning Co-Ordinator
Country: UK
Linkedin Profile: Yes

Re: Grid to Ground

Post by danielgadowski » Thu Jul 03, 2014 11:04 am

I believe that the problem stems back from where the point clouds were used originally - as-builds on industrial sites, where scale factor doesnt exist and you want everything 1:1.
For this reason people wanted "true" dimensions and were not bothered with scale.
Now, that the scanning has been adapted for many other uses, it does become a problem.
I agree, that an option for scaling should have been implemented into Cyclone a long time ago. Perhaps Cyclone 10? lol
Dan

geowac
V.I.P Member
V.I.P Member
Posts: 353
Joined: Thu Jan 23, 2014 10:07 am
Full Name: Stefan Wachsmann
Company Details: scantaxi.de
Company Position Title: supervisor
Country: Germany
Linkedin Profile: No
Location: Germany
Contact:

Re: Grid to Ground

Post by geowac » Thu Jul 03, 2014 9:30 pm

Interesting post!

We do the scaling after extracting the features and transform only the features.

In our case, the scaling of pointclouds has minor importance.

The question is something academic! I think, this has to be divided into two cases (may be more)...
  1. terrestrial scans with coverage of about 300m per scan
  2. arial (or long range scans) with distances of more than 500m from scanner to object (ground)
1) nothing should be done with the scanpoints because they (should !) represent the reality
2) refraction has to be taken into account

In both cases I guess, no scaling has to be used with the pointcloud, because the scale normally affects only the extracted features, projected to the map/ground. As long, as the pointcloud is used, no projection is needed (especially should not be needed). If extracted some features (the grid), it may be necessary to project them to the underlaying map/ground. I prefer a real projection instead of a scaling, because the scans provide real representation of the objects and scaling only could be distort the features.

If there is a scaling in the reference points - because of a coordinate transformation of some weird system - I would prefer to scale the extracted features instead of the cloud...

Even Orthophotos or something like this, extracted from the clouds, have to be transformed into the underlaying system. But I guess, a scaling wouldn't be the correct way. I think, it has to be a transformation instead...

If there are some different altitudes in the scans, that have to be considered, a projection would be somehow closer to the reality as a pure scaling - because you follow the direction of the projection in every point along gravity-normals or something like this...

In my opinion, in this case, scaling is only usable in flat systems like one map to another, but not 3D to 2D.

Regards
Stefan

User avatar
Jana-at-kubit
V.I.P Member
V.I.P Member
Posts: 236
Joined: Mon Aug 23, 2010 2:59 pm
Full Name: Jana Siebenbrodt
Company Details: FARO Technologie Inc [Dresden]
Company Position Title: Platform Owner
Country: Germany
Linkedin Profile: Yes
Location: Dresden, Germany
Contact:

Re: Grid to Ground

Post by Jana-at-kubit » Fri Jul 04, 2014 8:17 am

I absolutely agree to what Stefan wrote.
Important point on projection versus scaling!
Jana Siebenbrodt
----------------------------------------------------------
FARO Techologie Inc., Dresden, Germany

http://faro.com/
https://knowledge.faro.com/Software/
----------------------------------------------------------

User avatar
Phill
V.I.P Member
V.I.P Member
Posts: 586
Joined: Wed Feb 06, 2008 7:19 am
Full Name: Phillip Nixon
Company Details: SKM
Company Position Title: Surveyor
Country: Australia
Location: Sydney

Re: Grid to Ground

Post by Phill » Fri Jul 04, 2014 12:51 pm

Stefan

I will go into you post further at a later date, but answer me this, if I supply my client with point cloud on plane and data on grid, and he decides to measure an extra object from the cloud data and it's 900mm different do you think he will be upset. We live in a world were clients are now asking for cloud as well as extracted data.

We do a lot of work for departments of road who only work on grid. I believe a lot of people are using scanning on roads due to safety concerns.

Can't say much for aerial scams but if they are like I have seen all the data is scaled and projected to a grid system.

I will do the calcs on scale factor change due to height and refraction difference over my example 50m tommorrow when I find my textbook. The refraction one is a tough one as it changes with grazing lines etc, but I will just use the standard K factor.

Cheers
Phill

geowac
V.I.P Member
V.I.P Member
Posts: 353
Joined: Thu Jan 23, 2014 10:07 am
Full Name: Stefan Wachsmann
Company Details: scantaxi.de
Company Position Title: supervisor
Country: Germany
Linkedin Profile: No
Location: Germany
Contact:

Re: Grid to Ground

Post by geowac » Fri Jul 04, 2014 4:47 pm

Hi Phill,
Phill wrote:if I supply my client with point cloud on plane and data on grid, and he decides to measure an extra object from the cloud data and it's 900mm different do you think he will be upset. We live in a world were clients are now asking for cloud as well as extracted data.
This is a good point. Because the client will not have the background you have :(
The question is, what will be the correct way to handle the data.

If someone will measure the height of an Object (e.g. bridge) and notices that the clearing is wrong, because the scan is scaled, what will he mention? If you go outside and measure (or measure it in the original scan) the clearance would be fine! What do you do now? I don't know and I prefer to avoid this decisions.

The main problem is, that the grid is only 2.5D: 2D scaled (projected, but in the case of 2D this is nearly the same) and the marked altitudes are real world values, no scaling would be there!

To scale the scans only in two directions (x/y), may lead to wrong measurements, too (I did not calculate/verify this). The infrastructure planner will extract all objects in normal direction with real elevations. long measurements between your scan-points and the drawing will have some deviation, too. Because the scans follow the "bending" of the ground (this is a problem with totalstations and long distances, too). You have to correct this measurements. But if I am right, refraction depends on the distance. Is this the same as a scaling? Don't know, I have to reread about this (it's a long time ago I had this at university ;)

This will be the same problem with WebShare/TrueView, because I think, this data can't be scaled in any direction. If I extract some coordinates from this Views and compare the calculated distances with distances in the grid (drawing) they will be different too - but will match to the measurements in the unscaled scans...

Hope there is a mathematician here, who could explain this a little more!?

I always inform my clients, that the scans have a neighborhood accuracy but the grid has an over-all accuracy. Small measurements will be comparable, and elevations are correct over all (in the grid and scan). But sometimes it doesn't help but arguing about is easier.

By the way, if you make a vertical slice through the scaled pointcloud there will be a markable curve with the heights. How to handle this? I think, big pointcloud have to be deformed (transformed) to the grid, not only scaled!

Is an airborne scanner operator here, who can tell us how they do correct the data?

The ugly scans from the scanners are representing the real world, but the grid will be flat! I wished the Earth had stayed flat as 500 years ago :lol:

Cheers
Stefan

max72
V.I.P Member
V.I.P Member
Posts: 682
Joined: Tue Feb 26, 2013 9:32 pm
Full Name: Massimo De Marchi
Company Details: Massimo De Marchi
Company Position Title: freelancer
Country: Italy
Skype Name: massimo_de_marchi
Contact:

Re: Grid to Ground

Post by max72 » Fri Jul 04, 2014 9:36 pm

geowac wrote:Hi Phill,
I wished the Earth had stayed flat as 500 years ago :lol:
Easy:
http://theflatearthsociety.org/cms/
:-)
ing. Massimo De Marchi - +39 347 32 17 049 - www.studiodemarchi.net

Post Reply

Return to “Construction & Architecture”