TubeTK/Events/2010.10.26: Difference between revisions

From KitwarePublic
< TubeTK‎ | Events
Jump to navigationJump to search
(Created page with "= Danielle = * Computed weighting term between anisotropic and diffusive regularization = exponential decay as a function of distance to the closest polydata point * Registratio...")
 
No edit summary
 
(5 intermediate revisions by 3 users not shown)
Line 8: Line 8:


* Sliding tubes:
* Sliding tubes:
** Section of top tube moves 4 pixels to the right, section of bottom tube moves 4 pixels to the left


Original fixed image:
Original fixed image:
Line 34: Line 35:
[[File:SlidingBoxes_transformedMovingImage_anisotropic.png|500 px]]
[[File:SlidingBoxes_transformedMovingImage_anisotropic.png|500 px]]


* Sliding boxes:
* Sliding blocks:
** Top block moves 6 pixels to the right, bottom block moves 6 pixels to the left


Original fixed image:
[[File:SlidingBoxes2_originalFixedImage.png|500 px]]
Original moving image:
[[File:SlidingBoxes2_originalMovingImage.png|500 px]]


Organ borders:
[[File:SlidingBoxes2_border.png|500 px]]
Border normals:
[[File:SlidingBoxes2_normals.png|500 px]]
Motion field, no regularization:
[[File:SlidingBoxes2_resultingMotionField_noRegularization.png|500 px]]
Transformed moving image:
[[File:SlidingBoxes2_transformedMovingImage_noRegularization.png|500 px]]
Motion field, diffusive regularization:
[[File:SlidingBoxes2_resultingMotionField_diffusive.png|500 px]]
Transformed moving image:
[[File:SlidingBoxes2_transformedMovingImage_diffusive.png|500 px]]
Motion field, anisotropic diffusive regularization:
[[File:SlidingBoxes2_resultingMotionField_anisotropic.png|500 px]]
Transformed moving image:
[[File:SlidingBoxes2_transformedMovingImage_anisotropic.png|500 px]]
** Problem: corners within the fixed image do not have motion vectors, leading to problems in the transformed moving image
** Why? --> Because surface normals are discontinuous at corners...
+ Writing paper, due Thursday


* Writing paper, due Thursday
.


== Questions ==  
== Questions ==  
* Better way to find distance from a point to a polydata?
* Better way to find distance from a point to a polydata?
= Romain =
*  Implementation on the ComputeTortuosity module (only Distance Metric)
** debugging ++++++
*Future
** More debugging!
** #tube command
** Implementation of the other 2 methods : ICM SOAM
[[Category:TubeTK Events and Meetings|2010.10.26]]

Latest revision as of 18:49, 26 July 2013

Danielle

  • Computed weighting term between anisotropic and diffusive regularization = exponential decay as a function of distance to the closest polydata point
  • Registration works!


Sliding organ registration

  • Sliding tubes:
    • Section of top tube moves 4 pixels to the right, section of bottom tube moves 4 pixels to the left

Original fixed image: SlidingBoxes originalFixedImage.png Original moving image: SlidingBoxes originalMovingImage.png

Organ borders: SlidingBoxes border.png Border normals: SlidingBoxes normals2.png

Motion field, no regularization: SlidingBoxes resultingMotionField noRegularization.png Transformed moving image: SlidingBoxes transformedMovingImage noRegularization.png

Motion field, diffusive regularization: SlidingBoxes resultingMotionField diffusive.png Transformed moving image: SlidingBoxes transformedMovingImage diffusive.png

Motion field, anisotropic diffusive regularization: SlidingBoxes resultingMotionField anisotropic.png Transformed moving image: SlidingBoxes transformedMovingImage anisotropic.png

  • Sliding blocks:
    • Top block moves 6 pixels to the right, bottom block moves 6 pixels to the left

Original fixed image: SlidingBoxes2 originalFixedImage.png Original moving image: SlidingBoxes2 originalMovingImage.png

Organ borders: SlidingBoxes2 border.png Border normals: SlidingBoxes2 normals.png

Motion field, no regularization: SlidingBoxes2 resultingMotionField noRegularization.png Transformed moving image: SlidingBoxes2 transformedMovingImage noRegularization.png

Motion field, diffusive regularization: SlidingBoxes2 resultingMotionField diffusive.png Transformed moving image: SlidingBoxes2 transformedMovingImage diffusive.png

Motion field, anisotropic diffusive regularization: SlidingBoxes2 resultingMotionField anisotropic.png Transformed moving image: SlidingBoxes2 transformedMovingImage anisotropic.png

    • Problem: corners within the fixed image do not have motion vectors, leading to problems in the transformed moving image
    • Why? --> Because surface normals are discontinuous at corners...

+ Writing paper, due Thursday


Questions

  • Better way to find distance from a point to a polydata?


Romain

  • Implementation on the ComputeTortuosity module (only Distance Metric)
    • debugging ++++++
  • Future
    • More debugging!
    • #tube command
    • Implementation of the other 2 methods : ICM SOAM