Forum

Photogrammetric Opt...
 
Notifications
Clear all

[Sticky] Photogrammetric Optode Co-Registration with Cedalion

8 Posts
2 Users
0 Reactions
2,789 Views
Posts: 20
Admin
Topic starter
(@avl_openfnirsadm)
Eminent Member
Joined: 4 years ago
[#730]

This thread was created for discussions about Cedalion's photogrammetry-feature. An example notebook can be found here:

https://doc.ibs.tu-berlin.de/cedalion/doc/dev/examples/head_models/41_photogrammetric_optode_coregistration.html

A recorded "How-To" video from a recent NIRx webinar can be found here:
https://www.youtube.com/watch?v=upBmo2ZlWBM&feature=youtu.be


7 Replies
Posts: 4
(@olessia-jouravlevcarleton-ca)
New Member
Joined: 3 years ago

Hello, I tried implementing this code using our 3D image and snirf files from NIRx system. But in plot cluster step, the code recognizes only one green circle (optode) and do not recognize the others. We tried images using scaniverse and Structure 3D scanner, both had same issue. Is there a solution for this? Please let us know. 


Reply
Posts: 20
Admin
Topic starter
(@avl_openfnirsadm)
Eminent Member
Joined: 4 years ago

@olessia-jouravlevcarleton-ca without seeing your code this is a wild guess, but i wonder if you properly adjusted the color thresholds, which will depend on the sticker(color)s you use.

See https://doc.ibs.tu-berlin.de/cedalion/doc/dev/examples/head_models/41_photogrammetric_optode_coregistration.html#2.-Identifying-sticker-vertices

The example notebook uses yellow stickers and suggests how you can/should adapt the HSV values for your use case


Reply
1 Reply
(@olessia-jouravlevcarleton-ca)
Joined: 3 years ago

New Member
Posts: 4

@avl_openfnirsadm

Thank you for your instant reply. I am sharing my code here. The color thresholds are adjusted. My problem is it recognizes only one green circle. I have provided the screenshot for that as well. I wonder is taking scan in a 3D scanner matters? We are using scaniverse as mentioned in your workshop. 

Here is the link to the notebook :  https://drive.google.com/file/d/1mp6zTf6Nia_bucRosV59q4G4TbdGQnRq/view?usp=drive_link


Reply
Posts: 4
(@olessia-jouravlevcarleton-ca)
New Member
Joined: 3 years ago

Thank you for your instant reply. I am sharing my code here. The color thresholds are adjusted. My problem is it recognizes only one green circle. I have provided the screenshot for that as well. I wonder is taking scan in a 3D scanner matters? We are using scaniverse as mentioned in your workshop. 

Here is the link to the notebook : https://drive.google.com/file/d/1mp6zTf6Nia_bucRosV59q4G4TbdGQnRq/view?usp=drive_link

 


Reply
Posts: 20
Admin
Topic starter
(@avl_openfnirsadm)
Eminent Member
Joined: 4 years ago

@olessia-jouravlevcarleton-ca thanks for the notebook, but i could not find the screenshot of the scan. Could you share this (again) to provide a better intuition?

I think the 3D scanner does not matter, really, for this algorithm to work, if the resolution is somewhat decent. of course, precision might suffer for low res scans. The thing i am wondering is whether the units in your scan may be off, so that the cluster size for instance is off scale. in other words, if we are expecting 6mm clusters but in your scan the unit is off, that could explain something...

Also, can you try to run everything on the cedalion /dev branch? it might require you to reinstall the environment (more dependencies) but dev is by now light-years ahead of the last main release, the next major stable release is coming up within the next weeks!


Reply
1 Reply
(@olessia-jouravlevcarleton-ca)
Joined: 3 years ago

New Member
Posts: 4

@avl_openfnirsadm Hello Alex, I have uploaded the 3d image and snirf file in this folder. Please let me know if you can sort the problem out. https://drive.google.com/drive/folders/1eBfcRvuyYwAEm7S_Fd_ivIlFiw3OU71-?usp=drive_link

- Akshaya


Reply
Page 1 / 2
Share:
en_USEnglish