Scanstation P20 & Registration Accuracy

Trimble SX10, Leica BLK360, C5, C10, P15, P16, P20, P30, P40, P50, HDS8800, HDS8400, Riegl VZ6000, VZ4000, VZ2000, VZ1000 & VZ400i, Topcon GLS2000S, GLS1500, etc
User avatar
andrew.grigg
V.I.P Member
V.I.P Member
Posts: 170
Joined: Tue Aug 07, 2012 1:29 pm
Full Name: Andrew Grigg
Company Details: 40SEVEN
Company Position Title: Senior Land Surveyor
Country: England
Linkedin Profile: Yes

Re: Scanstation P20 & Registration Accuracy

Post by andrew.grigg » Fri Sep 19, 2014 1:02 am

Matt Young wrote:With the correct procedures you should achieve a mean average of around 1 or 2mm for your overall registration result. You must use the cloud 2 cloud function after the traverse in order to achieve this type of accuracy, and you must adjust constraints within the registration after the cloud 2 cloud process.

I always prefer using coincident targets with scans - as Paul suggests using no targets is fast but not reportable in the same way as traverse or coincident targets.

If you have a project with thousands of scans then don't try and do it all in one registration, it will be slow going. Try to break the project up into manageable areas. Two or three hundred scans max per registration is usually my preference.
Hi Matt,

Was about to post a question about this.

I'm working on a registration using the C10 which will eventually have around 300-400 scans. At the minute I have around 200 scans in a registration, and it is slow going when registering/making adjustments.

I recall you being able to add multiple registrations into 1 registration, but I've never managed to successfully do it.

I've tried to freeze the registration and create a ScanWorld, then add this to a new registration (so I could eventually do this with multiple registrations), but it never finds the constraints when Auto-Adding.

I haven't received the control yet for for the project, so am keen to ensure my scan registration is kept together.

Do you have any tips?

I also have a second query... I know that by cleaning data in the ControlSpace and then selecting "Set default Clouds", this will save the edits, but if your ScanWorlds are already part of a registration will they be applied when creating a ModelSpace, or do you need to re-add them to a registration?

Actually, I have a third query haha. If you have a registration which includes some cloud to cloud constraints, when you remove the ScanWorlds and re-add them, they disappear. Is there a way to prevent this happening, other than creating vertices in the MS and pasting in to the CS?

In response to the OP. I use the traverse function all the time on a C10 and it's great. Much quicker for me and great results. You can always set up additional targets over known positions and scan them in, and/or scan in temporary targets too, then add them all in to the registration.

User avatar
pburrows145
V.I.P Member
V.I.P Member
Posts: 1775
Joined: Wed Sep 12, 2007 9:16 am
Full Name: Paul Burrows
Company Details: Leica Geosystems Europe
Company Position Title: Scanning Solutions Manager - Europe
Country: UK
Linkedin Profile: Yes
Location: UK
Has thanked: 8 times
Been thanked: 5 times

Re: Scanstation P20 & Registration Accuracy

Post by pburrows145 » Fri Sep 19, 2014 6:30 am

I think you need to go on the official training course Andrew... Drop me a line to discuss if you wish. Ever thin you're asking here would be covered in some way.

User avatar
andrew.grigg
V.I.P Member
V.I.P Member
Posts: 170
Joined: Tue Aug 07, 2012 1:29 pm
Full Name: Andrew Grigg
Company Details: 40SEVEN
Company Position Title: Senior Land Surveyor
Country: England
Linkedin Profile: Yes

Re: Scanstation P20 & Registration Accuracy

Post by andrew.grigg » Fri Sep 19, 2014 9:59 am

Paul, I went on the course last year.

I've just checked the manual that is provided with the course and none of the topics I mentioned were covered.

I can't see any advice on working with massive registrations, merging registrations, or how to keep your cloud constraints between scanworlds when you remove them from a registration (if it's not possible that is ok!).

The course was good but shows basic registration techniques and cloud to clouding of around 10 scans, which in reality is nothing like a lot of the projects we work on.

User avatar
pburrows145
V.I.P Member
V.I.P Member
Posts: 1775
Joined: Wed Sep 12, 2007 9:16 am
Full Name: Paul Burrows
Company Details: Leica Geosystems Europe
Company Position Title: Scanning Solutions Manager - Europe
Country: UK
Linkedin Profile: Yes
Location: UK
Has thanked: 8 times
Been thanked: 5 times

Re: Scanstation P20 & Registration Accuracy

Post by pburrows145 » Fri Sep 19, 2014 10:06 am

Ok, well we offer bespoke courses for this type of thing.

Steven or Kate would run it, likely over 2 days at MK or your offices.

User avatar
Matt Young
Global Moderator
Global Moderator
Posts: 3884
Joined: Fri Nov 30, 2007 3:03 pm
Full Name: Matt Young
Company Details: Baker Hicks
Company Position Title: CAD-BIM Lead
Country: UK
Linkedin Profile: No
Has thanked: 34 times
Been thanked: 23 times

Re: Scanstation P20 & Registration Accuracy

Post by Matt Young » Mon Sep 22, 2014 9:56 am

As far as I know you cannot keep the cloud meshes if you remove scans from the registration.

You can create several registrations for one project and then add those registrations to additional registration.
If you don't see that there is nothing, then you are kidding yourself.

Jamesrye
V.I.P Member
V.I.P Member
Posts: 466
Joined: Mon Aug 11, 2008 4:13 pm
Full Name: James Rye
Company Details: Merrett Survey Partnership
Company Position Title: Spatial Analyst
Been thanked: 9 times

Re: Scanstation P20 & Registration Accuracy

Post by Jamesrye » Mon Sep 22, 2014 11:24 am

Andrew, I'm doing lots of projects with 2-300 scans or more, so perhaps I can offer some advice.

I'm working on a registration using the C10 which will eventually have around 300-400 scans. At the minute I have around 200 scans in a registration, and it is slow going when registering/making adjustments.

Try to separate the project into chunks - maybe by level or by independent control point placement.

I recall you being able to add multiple registrations into 1 registration, but I've never managed to successfully do it.

This should work if vertex's have corresponding ID's. If you are auto-finding targets from black and whites then Cyclone can add incorrect correspondences. I dont like adding multiple registrations into one and it shouldn't be necessary so long as you have a good control network from independent total station.

In response to the OP. I use the traverse function all the time on a C10 and it's great. Much quicker for me and great results. You can always set up additional targets over known positions and scan them in, and/or scan in temporary targets too, then add them all in to the registration.

True enough. I don't like the traverse function - yes it's quicker but I prefer to use resection. This can be quick too if you're auto-finding the targets. I then register the big lump of registered scans to independently surveyed control points done with the total station. This method provides easily understandable registration diagnostics. This goes without saying, but always try to work in a closed loops whenever possible.

And as Matt said, the cloud mesh dissapears if you remove it from the registration. I like cloud to cloud as a way to augment my work, but I'd never rely on it to join scans together without doing a lot of checking as you can get great looking registration diagnostics, but when you cut the sections, you see misalignment.

regards,

James

User avatar
Matt Young
Global Moderator
Global Moderator
Posts: 3884
Joined: Fri Nov 30, 2007 3:03 pm
Full Name: Matt Young
Company Details: Baker Hicks
Company Position Title: CAD-BIM Lead
Country: UK
Linkedin Profile: No
Has thanked: 34 times
Been thanked: 23 times

Re: Scanstation P20 & Registration Accuracy

Post by Matt Young » Mon Sep 22, 2014 12:02 pm

Thanks James for finding the words I couldn't on a Monday morning ;)
If you don't see that there is nothing, then you are kidding yourself.

markoscann
I have made <0 posts
I have made <0 posts
Posts: 5
Joined: Fri Nov 21, 2014 9:31 pm
Full Name: Marko Pejic
Company Details: University of Belgrade
Company Position Title: Assistant professor
Country: Serbia
Skype Name: marko.pejic77
Linkedin Profile: Yes

Re: Scanstation P20 & Registration Accuracy

Post by markoscann » Sat Nov 22, 2014 11:16 pm

Hi to all!
At Department of geodesy and geoinformatics in Belgrade, we owned P20 for almost a year and a half. I found that this peace of unit is breakthrough in the market, particularly compared to its Leica pulsed predecessors. But there is a significant drawback too, which is related to short range accuracy, on which I intend to discuss with you folks on the different occasion.
Traversing uses so called direct georeferencing method of every point cloud (known point occupation and backsighting). In short, total budget of errors using this procedure is followed by four parameter rigid body transformation model from internal coordinate system (CS) to field CS (3 axis translation and rotation around z axis), and includes instrumental errors (errors of the horizontal and vertical beam deflections, range finder errors and instrument leveling errors) + estimated 3D errors of occupied control points, centering and height measuring over instrument and target point and target recognition error. Mixed all together one can produce a covariance matrices of every resulted point in the point cloud.
Since the rotation of the system is performed only around Z axis, beside centering and height measuring error, the critical one could be the compensating error. P20 has internal devices that have survey grade accuracy. For example, the sensitivity of the dual axis compensator is top class and equals 1.5”. Measuring height with supplied tape is declared by 2 mm uncertainty, but I find the better performance. Also centering error depends on the instrument and backsight target height, and is declared about 1.5 mm for booth. Because of these “little things„ for this kind of TLS application, I think that Leica is leader among other manufactures, due to long tradition in producing surveying equipment. They just know all the tricks in surveying.
All this is a theory and assumptions, and here I will present some hard data. To prove the covariance model of the direct georeferencing using procedure of the known station and orientation, we make an experiment in the controlled environment of 5 ground control points for scanner setup, and different 6 on the pillars for accuracy estimation. Based on the sample of the 16 independent setups, we found that achieved georeferencing (registration) accuracy in the form of the 3D error vector is 1.8 mm in average and ranges from 0.5 mm to 4.0 mm. I have to mention that result contains the small and unsubstantial error component that comes from uncertainty of the estimated 3D position of pillars during verification of the conducted setup. Target recognition on pillars could have significant part of resulting error too.
One can conclude by himself, does P20 is appropriate for traversing… I found that P20 is in the class of a majority of total stations for this job. Plus, you can additionally use the surface matching or cloud constraints routines. Of course, backsighting range should be below 50-60 m, but not too short.
For my first post, I think that is quite enough :)

Post Reply

Return to “Pulse Based Scanners”