|
|
 | | From: | Ufit | | Subject: | 2D video into 3D space | | Date: | Fri, 21 Jan 2005 18:47:46 -0800 |
|
|
 | I was wondering if there are any algorithms converting f.ex. a sequence of video frames
into 3D objects. Or at least trying to do that. Are there any programs allowing that?
Thanks.
UFT
|
|
| | From: | Arthur J. O'Dwyer | | Subject: | Re: 2D video into 3D space | | Date: | Fri, 21 Jan 2005 22:25:00 -0500 (EST) |
|
|
|
On Fri, 21 Jan 2005, Ufit wrote:
>
> I was wondering if there are any algorithms converting f.ex. a sequence
> of video frames into 3D objects. Or at least trying to do that. Are
> there any programs allowing that? Thanks.
Please use lines of at most 75 characters (and preferably 65) on
Usenet. Re-formatting a badly broken posting takes time.
Do you mean that you have something like an MRI scan (or what passes
for such a scan in Hollywood movies;) and you want to turn it into a
solid object, in the reverse of the process that produced this video?
http://www.coppit.org/brain/topbottomslice.mpg
In that case, I'd suggest keywords "image segmentation" and "voxels",
or "edge detection" and "mesh generation". See for example
http://www.cs.berkeley.edu/~jrs/mesh/
Or do you mean that you have something like a video shot with a
video camera out of a car window, or a series of pictures of a rotating
object in space, and you want to reconstruct its shape by looking at
how features move in parallax? In that case, I'd suggest keywords
"feature detection [extraction]", "feature tracking", and "structure
from motion". See for example
http://homepages.inf.ed.ac.uk/rbf/CVonline/LOCAL_COPIES/OWENS/LECT12/node5.html
HTH,
-Arthur
|
|
| | From: | Ufit | | Subject: | Re: 2D video into 3D space | | Date: | Fri, 21 Jan 2005 22:03:56 -0800 |
|
|
|
"Arthur J. O'Dwyer" wrote in message news:Pine.LNX.4.60-041.0501212210060.18253@unix44.andrew.cmu.edu...
>
> Or do you mean that you have something like a video shot with a
> video camera out of a car window, or a series of pictures of a rotating
> object in space, and you want to reconstruct its shape by looking at
> how features move in parallax? In that case, I'd suggest keywords
> "feature detection [extraction]", "feature tracking", and "structure
> from motion". See for example
> http://homepages.inf.ed.ac.uk/rbf/CVonline/LOCAL_COPIES/OWENS/LECT12/node5.html
>
Yes - exactly the above one. Shape reconstruction was my interest.
It'd be nice to check it with some complete program or source code
I'm gonna research that. Thanks A.J.
U
|
|
|
