PDA

View Full Version : Convert 2D images to 3D images



ksudhakar100
03-19-2004, 10:33 PM
How to convert a set of 2D Images to 3D images,please list me some algorithms by which i can achieve this and the coding should be in C++

garf
03-21-2004, 05:01 AM
Originally posted by ksudhakar100:
How to convert a set of 2D Images to 3D images,please list me some algorithms by which i can achieve this and the coding should be in C++

http://www.opengl.org/discussion_boards/ubb/smile.gif
What do you mean?
Could you describe the geometry?
and the purpose of it?

nEUrOO
03-22-2004, 11:47 AM
Well ... i think you should calculate the delta-variation of color between pixels & then your third axe would be this variation.

(i think it sould be one method ;) )

maxyz
03-23-2004, 04:05 AM
is this a 3d texture ? if it is, you can use openGL 3d texture

eeeko
03-23-2004, 11:05 AM
Do you mean that you have, for example, 6 pictures, one from each angle showing the same picture and by "merging" those pictures you would like to have a 3d object pop out? Well, that doesn't sound like a simple thing to create. The best way would be that you should mark a point on each picture that his this point visible. By calculating the distance on each picture from, for example, the center of the picture you would get some kind of approximation to this. If their is a better method to achive this? I'am most certain of that! :-)

Good luck!

madmortigan
03-23-2004, 03:10 PM
If you're talking about getting the 3d model from various 2d pictures of it, search google for

3d model reconstruction from 2d image

In its general case, this is quite a problem. It gets easier, if you know what it is you're reconstructing.

madmortigan

ksudhakar100
03-23-2004, 08:01 PM
Hi All

The requirement is that there will be series of 2D images all these 2D images will be linked based on the meeting point and i have to reconstruct a 3D image from that .The 12 series of 2D images will be from MRI Scan and the format is DICOM format

JohnnyFowler
03-26-2004, 04:30 PM
Originally posted by ksudhakar100:
Hi All

The requirement is that there will be series of 2D images all these 2D images will be linked based on the meeting point and i have to reconstruct a 3D image from that .The 12 series of 2D images will be from MRI Scan and the format is DICOM format

JohnnyFowler
03-26-2004, 04:34 PM
Originally posted by JohnnyFowler:

Originally posted by ksudhakar100:
Hi All

The requirement is that there will be series of 2D images all these 2D images will be linked based on the meeting point and i have to reconstruct a 3D image from that .The 12 series of 2D images will be from MRI Scan and the format is DICOM formatSorry about that. I'm new here.

The standard method is called filtered back-projction. It involves convolution filtering or fft'ing of each image, then transforming each image to its original rotation position about a voxel set and back-projecting each pixel into the voxels. Sometimes alignment of the images is required using fiducial marks of known 3-d positions. You might check Chap. 3 of http://www.slaney.org/pct/pct-toc.html for more information.