The day after working with landowners and activists in Umm Salamuna in the West Bank, we stopped by to visit Alice Gray, who organized the protest. She runs a permaculture farm project in Bethlehem called Bustan Qaraaqa, and was interested in mapping the farm periodically to monitor growth and erosion. She was quite excited about the kite concept so we thought we’d give it a try.
The attempt was somewhat frustrating – while the day before we had over 30mph winds, at the farm we had the opposite problem: very little wind. BQ is located in a sharp valley where it’s hard to get a view from above (see picture) but we finally managed to loft a small parafoil kite (thanks Nadya!) with an iPod nano attached. We let it out about 500 feet, but the kite flew at a very low angle, and in the wrong direction. In the end we got a reasonable image of the other side of the valley, which we’ll try to rectify as a proof of concept.
We’re using VLC to review frames manually, saving the clearest ones as PNG images. Then we use hugin with the SIFT algorithm to auto-stitch the video frames together. This is a bit involved and if we continue using video we may want to make a web interface to do this automatically. As in: point it at a YouTube video and it generates a panorama as well as it can and opens it in Map Warper. As it turned out, the hilly terrain proved too much to successfully warp this capture into a usable map; see the final product. What we need is a much higher point of view, if there are going to be any hills. Luckily some of the kites I bought in San Juan last week are ideal for this – and they fly much more vertically so we won’t have to worry about being as far upwind of the target site.
These two days definitely show the ‘worst case scenario’ for this kind of mapping… super high and low winds, steep valleys and ridges, low-res video and time limitations. Anyways the best part of the day was when two Palestinian kids and their dad came out to see who the idiots were who couldn’t fly a kite… and helped us get one in the air:
Coming soon – our last day of mapping in the middle east was at the Royal Scientific Society of Jordan, where we captured some absolutely fantastic images under some of the best conditions we’ve seen so far.
Josh Levinger and I met up with some activists who were planting trees in Umm Salamuna (view in Google Maps) on a hillside which is scheduled to be annexed by a nearby Israeli settlement, and converted into a graveyard. The planting was organized by Alice Gray of Bustan Qaraaqa, so that if the land is taken over, the trees would have to be uprooted or chopped down before the land can be used.. As I understand it, one of the means by which settlements claim land is by using an Israeli law which opens land to new settlement if it has lain fallow for more than three years — so planting the hillside may defend it from such a claim.
The wind was so strong that our first kite, carefully made that morning from dowels and Tyvek, shattered immediately. Instead, we launched a small soft kite with an iPod nano attached to it. Here’s a stitched image of the video footage we captured:
See all the pictures on Flickr.
The iPod has an SD camera which can capture many hours of video – and it’s so super light that we can fly it on a pocket kite. Many of the frames are blurred and the resolution is pretty poor (we’d thought of using a Flip camera but they’re more expensive and heavier) but when you go through the footage frame by frame you can find lots of good images. We then stitched these together with Calico and got the above image. It helped a lot to put a small ‘sail’ on the back of the iPod so it didn’t spin as much.
Everyone was cold but once we started flying the kites we all got really excited. The owner of the land was there with his kids and they helped assemble the rig and fly the kite:
The mapping was a big success – everyone ‘got’ why we were doing it, that documenting the tree planting and how they’re changing the landscape is a form of testimony. We’re still working to rectify the imagery, and I’d like to ask folks if they have any ideas – the stitching software we’re using assumes images were taken from a single viewpoint, but the kite and camera were moving all over the place. As you can see above, the stitching distorts things and we lose a lot of detail – how can we reconstruct a high-res image that assumes multiple perspectives? I’m looking at this tutorial to start with. We’re also thinking about an algorithm to dump the clear, undistorted and unblurred frames from a movie file. Ideas?
Update: We’ll be adding this material to the Grassroots Mapping wiki, where we’re putting together a comprehensive guide on low-cost participatory mapping techniques. Our hope is that we can offer a Grassroots Mapping Kit which people can use to reproduce these techniques to explore and document their own geographies no matter where they are.
Cross-posted with the Center for Future Civic Media blog