Maurizio wrote:
>
> FLIGHT METADATA IS THE CRITICAL ISSUE: your final results will be as
> good as
> your flight metadata is.
>
> Don't ask: I have no idea how that data is actually collected :D

I can tell you a little bit about how it's done on the professional end - the 
plane's position is recorded with differential GPS, corrected against a base 
station ideally within about 50 miles. The orientation of the plane is recorded 
with an IMU (inertial measurement unit). Our contractor explained that the 
ownership of such units of high levels of accuracy is regulated in the USA, 
though I don't know the exact specifications or legal restrictions he was 
referring to. The camera is also mounted in a servo gimbal which not only keeps 
the camera oriented downwards, but can twist the camera to keep the frame 
oriented correctly if the plane is "crabbing" due to a sidewind, and also 
compensates for the forward motion of the plane by moving the camera backwards 
when the shutter is triggered. They have software for the GPS post-processing 
and another package which interpolates the position and orientation of the 
plane to the shutter trigger time, as well as doing the geometric computation 
to calculate the actual camera lens position (relative to the GPS antenna). As 
you might expect this is million dollar plus equipment.

I'm not an expert on what is available for UAVs, but I would expect you would 
be able to get GPS down to at submeter with some cleverness. Getting the 
orientation of the plane will be trickier - at a minimum you can get the 
approximate heading of the plane from the GPS log, but from what I've seen from 
the Camp Roberts UAV imagery it's not very reliable. You would probably want a 
digital compass and some sort of 3d gyro or accelerometer package to try to get 
the plane orientation. I suspect one of the challenges is not just getting high 
enough resolution at consumer prices, but also getting a fast enough response 
time. Most UAVs don't have as much mass as a Cessna and move around a lot more.

At the recent Random Hacks of Kindness event I was working with OAM listmember 
Jeffrey Johnson and Tim Caro-Bruce of Urban Mapping to create a system that 
would take roughly positioned UAV imagery and then using SIFT image matching  
would more precisely georeference each frame to existing orthoimagery. I have 
the code almost finished - if this turns out to be a wet weekend I'll hopefully 
get it done.

Marc

_______________________________________________
talk mailing list
[email protected]
http://openaerialmap.org/mailman/listinfo/talk_openaerialmap.org

Reply via email to