Discussion: Produce your own areal photo for mapping for only 200 $ !

I heard a newscast about this. Sounded intriguing! Basically you launch a helium balloon with a camera properly attached to it on a very long string, kind of like a kite.
The website is http://grassrootsmapping.org/

How's that better than 6 or 10 cm resolution photos we get for free? in most of the urbanized United States. What the non-professional enthusiast mappers like us at Get Lost!!, who have time to sit at the computer but not really the time to go out in the field, would like is access to algorithms for image and lidar processing... that would yield a product eminently usable for navigation without much manual massaging.

For $500 this would be more use. (and more fun)
http://www.youtube.com/watch?v=O0QJ063A04w
Then you add the Lidar...
http://www.terraluma.net/

Tundra, are you able to expand on what problems you have and what the algorithms and processing would do? Are there no algorithms available or just none available in the packages you're using?
I haven't done mapping but it is of interest to me especially the capture of highly detailed images and conversion into an orienteering map.

My problem is that everything that works costs $$$$$. What I would like to do is

* Vegetation edges from photos.
* Vegetation grades from photos based on user-assigned color masks/thresholds.
* Vegetation extraction out of raw .LAS lidar, plug and play.

All of this should produce nice Bezier vectors, not fragmented nor duplicate lines.

I would like a ratio of price to functionality at least on par with OCAD 10 Pro (€€€€).

I would think that orthorectification of the simple balloon-borne cameras would be difficult. You need good elevation data to do that - either from a stereo-pair itself or from an outside source. Making a stitched photo is one thing, but removing the distortions so you can make an accurate map from it is harder. Not that it can't be done, but it would be pretty involved. The lower the collection height and the steeper the terrain, the larger the topographic distortions will be. UAVs are the wave of the future for this type of data collection.

* Vegetation edges from photos.
I've discovered that the Mass 15/30cm series also have an NIR band available. I eventually want to play with that data to see if tree/field is easily discernible, or at least evergreen/deciduous

* Vegetation grades from photos based on user-assigned color masks/thresholds.
Already easy to do, not sure how useful it is though. I can setup a mapcalc program in GRASS to only keep data that has color data that matches some pattern, and then use that as a mask, but defining what those patterns are I think would be hard to do well.

* Vegetation extraction out of raw .LAS lidar, plug and play.
I think plug and play is always going to be hard with lidar. Maybe we can get to a state where there are a limited number of parameters to set though.

Log: enter it in the orienteering song thread.

In Manitoba they use balloons to make maps.

I doesn't meet Tundra/Desert's specifications, but OL Laser can be useful for creating a base map from lidar:

http://www.oapp.se/