| Re: Generate 3D Surface out of Points in Space [message #39807] |
Wed, 23 June 2004 10:21  |
Karl Schultz
Messages: 341
Registered: October 1999
|
Senior Member |
|
|
"Tukee10" <turgutkaracay@hotmail.com> wrote in message
news:b599c09a391226a337d5f4fe4e8672fd@localhost.talkaboutpro gramming.com...
> Hi everybody. I have 50 sliced images, that represent a splitting blood
> vessel. I went through the slices and extracted the contours of the blood
> vessel with the function CONTOURS. Now I have many points that lie on the
> surface of the vessel. I would like to reconstruct the 3d surface of the
> vessel out of these points. Is there any function or procedure that is
> able to do that?
Surface reconstruction is a hard problem. If you google around you will see
that quite a few people have spent much of their lives working on small
parts of the problem.
That being said, if your data has certain constraints, then the job can be
done.
In your case, are the images such that the blood vessels are a very distinct
color, or otherwise distinguishable from the rest of the image? If so, you
can stack the images into a volume and then run ISOSURFACE to generate a
surface. You would need to determine what isovalue to use that selects the
pixels representing the vessels the best. You might consider running some
image filtering tools to sharpen or clean up the images to make it as close
to "on-off" as possible with respect to the blood vessel pixels.
I think that this is your best shot, given what I know about your data.
Karl
|
|
|
|
|
|
|
|
| Re: Generate 3D Surface out of Points in Space [message #39945 is a reply to message #39807] |
Thu, 24 June 2004 09:33  |
siliconcube
Messages: 11
Registered: June 2004
|
Junior Member |
|
|
I had the same problem and this board did a great job helping me
figure out where i went wrong. I had 85 SEM image slices, instead of
having contours I just cleared the background noise leaving me with
whatever it is I was looking at. Then I reduced resolution (otherwise
IDL would give me problems with lack of memory and i have a 5GB of DDR
Ram installed), my image files went from 1.8MB each to about 60KB each
and i ran the code (slightly modified) from Dr. Fanning's website
which he posted in the earlier message and everythign worked out
pretty darn well =D.
Good luck
Aleks
"Karl Schultz" <kschultz_no_spam@rsinc.com> wrote in message news:<10djeuf4dhht95b@corp.supernews.com>...
> "Tukee10" <turgutkaracay@hotmail.com> wrote in message
> news:b599c09a391226a337d5f4fe4e8672fd@localhost.talkaboutpro gramming.com...
>> Hi everybody. I have 50 sliced images, that represent a splitting blood
>> vessel. I went through the slices and extracted the contours of the blood
>> vessel with the function CONTOURS. Now I have many points that lie on the
>> surface of the vessel. I would like to reconstruct the 3d surface of the
>> vessel out of these points. Is there any function or procedure that is
>> able to do that?
>
> Surface reconstruction is a hard problem. If you google around you will see
> that quite a few people have spent much of their lives working on small
> parts of the problem.
>
> That being said, if your data has certain constraints, then the job can be
> done.
>
> In your case, are the images such that the blood vessels are a very distinct
> color, or otherwise distinguishable from the rest of the image? If so, you
> can stack the images into a volume and then run ISOSURFACE to generate a
> surface. You would need to determine what isovalue to use that selects the
> pixels representing the vessels the best. You might consider running some
> image filtering tools to sharpen or clean up the images to make it as close
> to "on-off" as possible with respect to the blood vessel pixels.
>
> I think that this is your best shot, given what I know about your data.
>
> Karl
|
|
|
|
| Re: Generate 3D Surface out of Points in Space [message #39946 is a reply to message #39817] |
Thu, 24 June 2004 00:57 |
Beat.Schmutz
Messages: 6
Registered: October 2003
|
Junior Member |
|
|
> I tried your method of creating a mesh structure out of MRI slices before,
> but it failed, because the ComputeMesh-method doesn`t work as soon as you
> get two contours on one plane.
>
I had the same problem, so I used SHADE_VOLUME instead. I filled the
extracted contours with black (0), created a volume with the stacked
contours and then set VALUE and LOW in SHADE_VOLUME to zero. This
worked well for the 3D reconstruction of a bone (distal Femur)
surface. I don't know whether it will work in your case.
Cheers,
Beat
|
|
|
|
comp.lang.idl-pvwave archive
Members
Search
Help
Login
Home
