[ptx] Optimizing, Stitching and Seaming

Pablo d'Angelo pablo at mathematik.uni-ulm.de
Mon Nov 24 17:45:30 GMT 2003


On Mon, 24 Nov 2003, Yili Zhao wrote:

> On Mon, 24 Nov 2003, Pablo wrote:
> > No need for 3d space. We don't care how far away the pixels are, since we
> > just have one viewpoint where all view rays intersect.
>   Do not understand. I can get a transform matrix for every image, but how 
> can I construct a panorama from these images and the associated matrix? 
> May you interpret it more in detail?

this is how the mapping pano -> image coordinates is done in panotools:

The point is always twodimensional. x,y for rectilinear images, or yaw and
pitch for equirect. images and so on.

 1. pano_proj -> erect
 2. rotate equirect?? ( rotate yaw in screenpoints ? )
 3. spherical -> erect ??
 4. persp_sphere ( rotate point with rotation
                   matrix)
 5. sphere_tp -> image_proj
 6. distortion correction
 7. shift image origin
 8. shear image

I'm not sure what steps 2 and 3 actually do, I would have to look at the
sourcecode of pano tools to figure that out. You can look at the sourcecode of
Panotools and read the rest there, though its a bit tedious to do.

ciao
  Pablo
--
http://wurm.wohnheim.uni-ulm.de/~redman/
Please use PGP


More information about the ptX mailing list