Scanning railways

Please post any tips or advice you have in general relating to laser scanning either whilst onsite or office based.
Post Reply
MalteHC
V.I.P Member
V.I.P Member
Posts: 144
Joined: Mon May 05, 2014 11:19 am
Full Name: Malte Holm
Company Details: TLS
Company Position Title: Surveyor
Country: Norway
Linkedin Profile: Yes

Scanning railways

Post by MalteHC » Wed Sep 03, 2014 10:08 am

Hi guys

I am in the starting process of planning a project.
The project is about finding a automatic way of calculating if there is to many or few chippings between the rail lines.
The data is collected with a scanner, and is orientated. So the only worries is how to calculate the chippings.
But what is nagging me, is that there isn't a real point zero to calculate the surface model from. My first thoughts is to make a triangular surface model and somehow use the rail lines as a point of zero. But i can't figure out how to extract the rail lines from the rest of the scanning? Even more, the chip pings kind of floats into the outside grass, so i have to figure out a method, how to calculate an outline of the railways triangular surface model.

I am thinking of selecting the rail lines, by somehow use a sort of polynomial?

But i can't figure out how to calculate an outline.

Does anyone have some experience or an idea, that can help to solve some of my problems?

User avatar
richard_m
V.I.P Member
V.I.P Member
Posts: 280
Joined: Thu Apr 03, 2014 9:14 pm
Full Name: Richard Merrin
Company Details: Fugro
Company Position Title: Data Processor _ Project Engineer
Country: Netherlands
Linkedin Profile: Yes
Contact:

Re: Scanning railways

Post by richard_m » Wed Sep 03, 2014 10:51 am

Hi,

it is normal to relate all measurements to the theoretical centreline between the 2 rails. We've been busy with R&D in this area for a while, I cannot say much unfortunately because:

1 - this is sensitive i.p. for us
2 - I'm too stupid to reliably know

Be aware though, if this is a research project we are already doing this commercially..

http://www.fugro.nl/related/brochures-p ... ailmap.pdf

Regards

Richard

MalteHC
V.I.P Member
V.I.P Member
Posts: 144
Joined: Mon May 05, 2014 11:19 am
Full Name: Malte Holm
Company Details: TLS
Company Position Title: Surveyor
Country: Norway
Linkedin Profile: Yes

Re: Scanning railways

Post by MalteHC » Wed Sep 03, 2014 11:58 am

Thanks for your reply. But it still leaves me asking, how i can define the two rail lines and how is it possible to define an outliner?
although it is a research project and i know that a lot of companies can provide this service, but unfortually the math in their software is a black box and i need to know all this basics, to research further into this area.

mjankor
V.I.P Member
V.I.P Member
Posts: 224
Joined: Thu Oct 15, 2009 12:51 pm
Full Name: Martin Ankor
Company Details: Custom Spatial Solutions
Company Position Title: Principle
Country: Australia

Re: Scanning railways

Post by mjankor » Wed Sep 03, 2014 3:28 pm

MalteHC wrote:Hi guys

I am in the starting process of planning a project.
The project is about finding a automatic way of calculating if there is to many or few chippings between the rail lines.
The data is collected with a scanner, and is orientated. So the only worries is how to calculate the chippings.
But what is nagging me, is that there isn't a real point zero to calculate the surface model from. My first thoughts is to make a triangular surface model and somehow use the rail lines as a point of zero. But i can't figure out how to extract the rail lines from the rest of the scanning? Even more, the chip pings kind of floats into the outside grass, so i have to figure out a method, how to calculate an outline of the railways triangular surface model.

I am thinking of selecting the rail lines, by somehow use a sort of polynomial?

But i can't figure out how to calculate an outline.

Does anyone have some experience or an idea, that can help to solve some of my problems?
Maybe use the highest points (filter them) to define the rail lines, and also, therefore, extents of the chipping analysis.

ucab127
I have made <0 posts
I have made <0 posts
Posts: 9
Joined: Fri Feb 25, 2011 12:18 pm
Full Name: Anita Soni
Company Details: UCL
Company Position Title: Research Engineer
Country: UK
Linkedin Profile: Yes

Re: Scanning railways

Post by ucab127 » Wed Sep 03, 2014 5:04 pm

I'm not sure if this is what you're looking for/how helpful this will be for you but there's an academic paper from an ISPRS conference on the automatic extraction of rail track (generic) using various algorithms based on the height of the rail to segment it from the rest of the point cloud from mobile scanning. It references a couple of MSc dissertations which are worth reading if you want to understand more about the extraction process:

http://www.isprs-ann-photogramm-remote- ... 3-2013.pdf

MalteHC
V.I.P Member
V.I.P Member
Posts: 144
Joined: Mon May 05, 2014 11:19 am
Full Name: Malte Holm
Company Details: TLS
Company Position Title: Surveyor
Country: Norway
Linkedin Profile: Yes

Re: Scanning railways

Post by MalteHC » Thu Sep 04, 2014 11:24 am

mjankor wrote:
MalteHC wrote:Hi guys

I am in the starting process of planning a project.
The project is about finding a automatic way of calculating if there is to many or few chippings between the rail lines.
The data is collected with a scanner, and is orientated. So the only worries is how to calculate the chippings.
But what is nagging me, is that there isn't a real point zero to calculate the surface model from. My first thoughts is to make a triangular surface model and somehow use the rail lines as a point of zero. But i can't figure out how to extract the rail lines from the rest of the scanning? Even more, the chip pings kind of floats into the outside grass, so i have to figure out a method, how to calculate an outline of the railways triangular surface model.

I am thinking of selecting the rail lines, by somehow use a sort of polynomial?

But i can't figure out how to calculate an outline.

Does anyone have some experience or an idea, that can help to solve some of my problems?
Maybe use the highest points (filter them) to define the rail lines, and also, therefore, extents of the chipping analysis.

Thanks, i think this idea is worth investigating further into.

Jennifer L.
I have made 60-70 posts
I have made 60-70 posts
Posts: 62
Joined: Fri Dec 06, 2013 2:02 pm
Full Name: Jennifer Ludwig
Company Details: Technodigit
Company Position Title: Application Engineer
Country: France
Linkedin Profile: Yes
Contact:

Re: Scanning railways

Post by Jennifer L. » Fri Sep 05, 2014 10:32 am

Hi,
3DReshaper Application has a plugin which allows to write Scripts in order to run automatic processing and also to write your own process.
A Script extracting automatically the railways from point clouds has been written with 3DReshaper for a French company working for French railroad.
http://www.3dreshaper.com/en1/En_script.htm
Unfortunately, it is not possible to find more "public" information on the internet. So if you are interested, please contact us and we will be able to show you a demo of a basic script extracting rails. We can't give you the code itself, of course.
Jennifer

User avatar
richard_m
V.I.P Member
V.I.P Member
Posts: 280
Joined: Thu Apr 03, 2014 9:14 pm
Full Name: Richard Merrin
Company Details: Fugro
Company Position Title: Data Processor _ Project Engineer
Country: Netherlands
Linkedin Profile: Yes
Contact:

Re: Scanning railways

Post by richard_m » Fri Sep 05, 2014 10:56 am

Probably worth remembering that each country only has a handful of types of rails and accurate profile diagrams of these are readily available. Normally you can read the type of rails from a code cast onto the side but of course in reality you can't stop a "scanning train" on live rails to have a look, for academic purposes you can probably assume the type of rails is known. Even if not the scan data simply needs fitting to a "parts table" containing the profiles of the rail types in order to ID the type and then it should be realitively easy to "follow" the rail in the data. This will allow you to ascertain the geometry of the rails and reduce this to the theoretical track centreline. A few mm of wear is likely on the rails but if its only the ballast bed you're interested then you have a couple of cm to play with.

I'm assuming you have existing data and scanned from above the track.

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: Scanning railways

Post by danielgadowski » Mon Sep 08, 2014 1:06 pm

Hi there,
i'll throw my two cents here as a person who actually works on track.
1. I would aproach such project with a trolley kinematic scanning system in mind (Amberg GRP5000).
2. The trolley would measure your track, the scanner would pick up your balast.
3. Extract profiles from your data set. These will have the track rinning edhes on them plus a profile through the scan.
4. Use the known offsets from your running edges to get your "sleepers base surface".
5. Use the profiles to create the second surface
6. Calculate the volume.
I hope this helps.
Daniel

User avatar
richard_m
V.I.P Member
V.I.P Member
Posts: 280
Joined: Thu Apr 03, 2014 9:14 pm
Full Name: Richard Merrin
Company Details: Fugro
Company Position Title: Data Processor _ Project Engineer
Country: Netherlands
Linkedin Profile: Yes
Contact:

Re: Scanning railways

Post by richard_m » Mon Sep 08, 2014 2:07 pm

Malte,

I assumed you were working from existing scan data, if not how is the track scanned? From a vehicle/train or trolley or from static scans on the trackside etc?

Richard.

Post Reply

Return to “Tips & Advice”