[ptx] autopano: first attempt at semi-automatic panorama stitching

[Yili Zhao]

ED wrote:
>if it was only the initial overlap which was the problem 
>then I suspect that most users would be ecstatic if they merely had to 
>visually drag the pictures into rough alignment and let the auto 
>algorithm do the rest!  PTGui has a nice mode which lets you do this (as 
>an example of an implementation).
I aslo like that feature and think it is essential to initial parameters
estimation, and it should be added in hugin wishlist for future implementation.

Pablo wrote:
>The Fourier mellin transform for example.
you can see Fourier mellin transfrom matlab code from 
http://www.ecs.soton.ac.uk/~aiw99r/fm_transform/

For auto pano and initial params estimation, I have some notes:
paper[2] said "To handle larger initial displacements, a hierarchical coarse-to-fine
optimization scheme can be used". From below URL 
http://www.cs.washington.edu/education/courses/576/03sp/projects/project2/project2.htm
you can dowload a skeleton code (VC6 and VC7) and a sample executable program. It seems
that the so called "hierarchical coarse-to-fine" is a Lucas-Kanade distortion.

Pablo, since you have some experience with KLT, if you have time, maybe you can read the
papers and the pages roughly and give some comments.

Reference:
[1] J.R.Bergen,P.Anandan, K.J.Hanna,and R.Hingorani. Hierarchical mode-based motion estimation. In Second European Conference on Computer Vision (ECCV'92),pages 237-252,May 1992. (I can not find this paper)
[2] R.Szeliski and H.-Y.Shum. Creating full view panoramic image mosaics and environment maps.In Computer Graphics
(SIGGRAPH'97) Proceedings, pages 251-258, August 1997.
[3] H.Shum and R.Szeliski. Construction and refinement of panoramic mosaics with global and local alignment. In Proc.
Intl. Conf. on Computer Vision, 1998.

Regards
Yili