I had some thoughts on the kite mapping. It seems that the hard part
is the image rectification. The way the grassrootsmapping [1] guys
are doing it is manually relative to a map or some other aerial
imagery. While thinking about how you could automate it, the three
things that came to mind are geotagging the photos, having reference
points on the ground, and the UW photo tourism project.
For geotagging [2] the photos, you'd need to attach a gps receiver, or
data logger near the camera, so that would increase the payload
weight, and it still doesn't give you pointing direction, although you
might be able to add that after the fact by hand (which could be
tedious). The geotags might lessen the computations to rectify the
photos.
The reference points on the ground would be fairly easy. Just have a
banner laying on the ground with a symbol large enough to be seen in
the photos, and note its coordinates with a gps unit. Garmin has a
way to average the coordinates of a waypoint after you've just made
it, but I haven't tested it to see how much accuracy it adds. I
suppose the best way to test that is find the coordinates of a
benchmark, and compare them after averaging. You'd probably want at
least three in the closest area, and maybe a few farther out. Then
you could use your own coordinates to rectify the photos, rather than
pulling them off other imagery, or maps.
The photo tourism project [3] is probably the neatest idea, although I
haven't tried it, and they imply that it's quite computationally
intensive. It takes a lot of photos and uses them to compute a 3-d
point cloud of the object being photographed. I don't know how well
it would do on relatively flat subjects though. They've posted some
source code so that others can try it out. It's one of the
technologies that went into Microsoft's Photosynth [4].
It certainly does sound like a fun and easy way to get aerial imagery,
but some work needs to be done to make the processing less labor
intensive, and reliant on outside sources for the rectification. It
may be paranoia on my part, but it sounds to me like a claim could be
made for the rectified kite imagery being a derivative work of the
rectification source (e.g. google satellite view).
Ramey
[1]
http://grassrootsmapping.org/
[2]
http://en.wikipedia.org/wiki/Geotagging
[3]
http://phototour.cs.washington.edu/
[4]
http://photosynth.net/