new hsi/hpi script to generate control points from prealigned images

[kfj]

Hi group!

I've spent the last weeks working on a large script using hsi/hpi, the
hugin scripting/plugin interface. Now I've put the first version
online, together with a README file, at my bazaar repo:

http://bazaar.launchpad.net/~kfj/+junk/script/view/head:/main/woa.py
http://bazaar.launchpad.net/~kfj/+junk/script/view/head:/main/README.woa

woa stands for 'warped overlap analysis'. To give you an idea what
it's all about, I'll quote my own readme:

The method is easily explained: for each pair of overlapping images,

- Take the overlap as your target area.

- Choose a projection where this area is quite undistorted.

- Calculate the partial image for each image that corresponds
to the target area, using the same projection (warp).

- Once you have these warped overlap images, they should be quite
similar and undistorted. Therefore they are good source material
for a CPG. Use the CPG on the warped overlap images.

- Transform the control points you have obtained back to the
original image coordinates.

I suppose the approach isn't new, but I've tried to be meticulous
about it and implement it to high precision. This script won't run
without hsi/hpi and needs a few other things which aren't quite
standard, but easy to get. You can get hsi/hpi by compiling hugin's
python_scripting branch with BUILD_HSI=ON. This works hopefully on
Linuxes and Windows, Mac OS isn't supported. See the readme for more
detailed information. The intended use is mainly mixed-lens panoramas
and lens calibration. Comments welcome!

Kay