Our client 3D Welt Vermessung (https://3dwelt-vermessung.de/) just started a series of similar sized projects of critical infrastructure in Germany. It is featured in the Intergeo-issue of GIM-International. The article can be found from page 19 onwards.
https://www.gim-international.com/magazines/gim-issue-6
Cheers
Daniel
Documenting a medical clinic with 12932 scans...
- Daniel Wujanz
- V.I.P Member
- Posts: 227
- Joined: Fri Aug 24, 2018 11:26 am
- 5
- Full Name: Daniel Wujanz
- Company Details: technet GmbH
- Company Position Title: 3D Laser Scanning Specialist
- Country: Germany
- Linkedin Profile: Yes
- Location: Berlin
- Has thanked: 206 times
- Been thanked: 188 times
- Contact:
- landmeterbeuckx
- V.I.P Member
- Posts: 1615
- Joined: Tue May 01, 2012 5:19 pm
- 11
- Full Name: Lieven Beuckx
- Company Details: Studiebureau Beuckx
- Company Position Title: Owner
- Country: Belgium
- Linkedin Profile: Yes
- Has thanked: 183 times
- Been thanked: 548 times
Re: Documenting a medical clinic with 12932 scans...
Now that's a scanning project for sure. Great read.
-
- V.I.P Member
- Posts: 544
- Joined: Fri Feb 24, 2017 10:47 am
- 7
- Full Name: Martin Graner
- Company Details: PointCab GmbH
- Company Position Title: Research and Development
- Country: Germany
- Linkedin Profile: No
- Has thanked: 160 times
- Been thanked: 175 times
- Contact:
Re: Documenting a medical clinic with 12932 scans...
Nearly 13,000 scan positions.Daniel Wujanz wrote: ↑Mon Oct 31, 2022 8:38 am Our client 3D Welt Vermessung (https://3dwelt-vermessung.de/) just started a series of similar sized projects of critical infrastructure in Germany. It is featured in the Intergeo-issue of GIM-International. The article can be found from page 19 onwards.
https://www.gim-international.com/magazines/gim-issue-6
Cheers
Daniel
Daniel is not impressed
3 weeks of registering.
I don't wanna know how long they are going to model this thing....
You do not have the required permissions to view the files attached to this post.
- Daniel Wujanz
- V.I.P Member
- Posts: 227
- Joined: Fri Aug 24, 2018 11:26 am
- 5
- Full Name: Daniel Wujanz
- Company Details: technet GmbH
- Company Position Title: 3D Laser Scanning Specialist
- Country: Germany
- Linkedin Profile: Yes
- Location: Berlin
- Has thanked: 206 times
- Been thanked: 188 times
- Contact:
Re: Documenting a medical clinic with 12932 scans...
Oh, yes I was : ) We see projects between 4000 to 6000 scans on a regular basis, but this thing was a different breed.Nearly 13,000 scan positions.
Daniel is not impressed
In fact, we had to assign some more memory to our adjustment which did the split once you hit around 12000 pairwise registrations. In turns out, now you go mental...
This project will go on for a long time. As far as I understood they will start modelling components once they are requested.3 weeks of registering.
I don't wanna know how long they are going to model this thing....
All the best
Daniel
- gord
- I have made 20-30 posts
- Posts: 29
- Joined: Tue Dec 22, 2020 9:30 pm
- 3
- Full Name: Gord Anderson
- Company Details: tiyuti
- Company Position Title: developer
- Country: Australia
- Linkedin Profile: No
- Has thanked: 39 times
- Been thanked: 10 times
Re: Documenting a medical clinic with 12932 scans...
Fascinating article .. particularly auto detecting planes and matching them to help register such a vast number of scans !Daniel Wujanz wrote: ↑Mon Oct 31, 2022 8:38 am Our client 3D Welt Vermessung (https://3dwelt-vermessung.de/) just started a series of similar sized projects of critical infrastructure in Germany. It is featured in the Intergeo-issue of GIM-International. The article can be found from page 19 onwards.
https://www.gim-international.com/magazines/gim-issue-6
I often wondered if matching line segments found in horizontal slices of each scan, would be a good way to do auto-registration for buildings - or at least reduce computation, as a first step.
Thanks for sharing this.
ps : web page of the same article here : https://www.gim-international.com/conte ... n-planning
gord anderson
quato.xyz
quato.xyz
- Daniel Wujanz
- V.I.P Member
- Posts: 227
- Joined: Fri Aug 24, 2018 11:26 am
- 5
- Full Name: Daniel Wujanz
- Company Details: technet GmbH
- Company Position Title: 3D Laser Scanning Specialist
- Country: Germany
- Linkedin Profile: Yes
- Location: Berlin
- Has thanked: 206 times
- Been thanked: 188 times
- Contact:
Re: Documenting a medical clinic with 12932 scans...
G'day Gord,
thanks for the flowers! There are some scientific approaches that make use of detected 3D-lines for registration - yet, they were published in German. If you're up to extreme challenges then have a look at the second reference at the very bottom.
The disadvantages of using lines are: a) you don't use as many points of the orignal data in comparison to planar patches and b) if there are no edges in the scene, then there will be no 3D-lines. We have many clients in the ship building industry where you don't see many edges but a lot of small planar patches on the curved ship hull. However, I'm sure there are some fields of application where they could be handy.
The computational load of matching planes is not very demanding. However, you could save some time and memory when just loading a slice for edge detection. Which brings us to the problem of extracting slices. If you'd be using a structured point cloud format, then this is a fairly quick operation. If you'd be sporting unstructured data I reckon detecting planes (or other geometric primitives) rather than line segments makes more sense since the entire point cloud has to be entirely loaded anyway.
Horizontal sections are used in some "field registration" / pre-registration solutions. The quality is definetely sufficient in most cases. Hence, it does not make too much sense to develop a solution which gives you more accurate starting values for registration algorithms that compute the final alignment of the point clouds.
Here's my list of arguments against auto-registration based on edges for buildings : )
- Z+F developed the first scanner that provided pre-orientation of captured scans (2015)
- Riegl followed just a few month later (2015)
- Leica provided a solution with the RTC to the same issue in 2018
- ...
In other words, if you could solve this issue purely based on algorithms, why would Z+F, Riegl, Leica, Trimble and Faro be so stupid do invest heaps of money to develop a pre-orientation solution?
The enemies of every registration algorithm are combinatorics (you don't want to compute all possible ones since live is short) and symmetry. The aforementioned sensor-based solutions help you to drastically reduce the amount of potential registrations and you will match e.g. identical rooms that are located on different levels.
Cheers and thanks again
Daniel
P.S. Scantra goes back to this publication from 2004: https://www.fig.net/resources/proceedin ... _et_al.pdf
https://gispoint.de/artikelarchiv/avn/2 ... tiven.html
thanks for the flowers! There are some scientific approaches that make use of detected 3D-lines for registration - yet, they were published in German. If you're up to extreme challenges then have a look at the second reference at the very bottom.
The disadvantages of using lines are: a) you don't use as many points of the orignal data in comparison to planar patches and b) if there are no edges in the scene, then there will be no 3D-lines. We have many clients in the ship building industry where you don't see many edges but a lot of small planar patches on the curved ship hull. However, I'm sure there are some fields of application where they could be handy.
The computational load of matching planes is not very demanding. However, you could save some time and memory when just loading a slice for edge detection. Which brings us to the problem of extracting slices. If you'd be using a structured point cloud format, then this is a fairly quick operation. If you'd be sporting unstructured data I reckon detecting planes (or other geometric primitives) rather than line segments makes more sense since the entire point cloud has to be entirely loaded anyway.
Horizontal sections are used in some "field registration" / pre-registration solutions. The quality is definetely sufficient in most cases. Hence, it does not make too much sense to develop a solution which gives you more accurate starting values for registration algorithms that compute the final alignment of the point clouds.
Here's my list of arguments against auto-registration based on edges for buildings : )
- Z+F developed the first scanner that provided pre-orientation of captured scans (2015)
- Riegl followed just a few month later (2015)
- Leica provided a solution with the RTC to the same issue in 2018
- ...
In other words, if you could solve this issue purely based on algorithms, why would Z+F, Riegl, Leica, Trimble and Faro be so stupid do invest heaps of money to develop a pre-orientation solution?
The enemies of every registration algorithm are combinatorics (you don't want to compute all possible ones since live is short) and symmetry. The aforementioned sensor-based solutions help you to drastically reduce the amount of potential registrations and you will match e.g. identical rooms that are located on different levels.
Cheers and thanks again
Daniel
P.S. Scantra goes back to this publication from 2004: https://www.fig.net/resources/proceedin ... _et_al.pdf
https://gispoint.de/artikelarchiv/avn/2 ... tiven.html
- smacl
- Global Moderator
- Posts: 1409
- Joined: Tue Jan 25, 2011 5:12 pm
- 13
- Full Name: Shane MacLaughlin
- Company Details: Atlas Computers Ltd
- Company Position Title: Managing Director
- Country: Ireland
- Linkedin Profile: Yes
- Location: Ireland
- Has thanked: 627 times
- Been thanked: 657 times
- Contact:
Re: Documenting a medical clinic with 12932 scans...
Fantastic article Daniel. I still think you have the makings of a great reference book for both practitioners and academics if you were to assemble the material you've put out there.
- Daniel Wujanz
- V.I.P Member
- Posts: 227
- Joined: Fri Aug 24, 2018 11:26 am
- 5
- Full Name: Daniel Wujanz
- Company Details: technet GmbH
- Company Position Title: 3D Laser Scanning Specialist
- Country: Germany
- Linkedin Profile: Yes
- Location: Berlin
- Has thanked: 206 times
- Been thanked: 188 times
- Contact:
Re: Documenting a medical clinic with 12932 scans...
Cheers Shane!
I wrote a book chapter plus a guide to registration over the cause of the last two years. I wouldn't even dare how long it may take to finish off an entire book. Long story short: That's officially too much work : )
All the best
Daniel
I wrote a book chapter plus a guide to registration over the cause of the last two years. I wouldn't even dare how long it may take to finish off an entire book. Long story short: That's officially too much work : )
All the best
Daniel