how will it be screened in terms of accuracy?
My third point "who will preprocess/verify data?" was meant to cover that.
But I am glad you focussed on the issue. I make it a subject "QA- and who is
to fix it ?"
Too many take for granted that aerial imagery is 100% accurate and it's just
matter of pushing a magic button to get it served.
The fact is that at least in my part of the world that simply does not
happen and my experience with material from other continents has confirmed
this again and again.
After all aerials it should be kept in mind that aerials are supplied by
contractors who employ humans who every now and then get distracted (or take
shortcuts to meet deadlines).
In a my fantasy world, I think it would be neat to be able to connect to a
tile server and pass parameters defining accuracy,
recency, resolution or whatever, and get back matching tiles.
Uhmm ... I am a believer that anything is possible.
Still some 'perfect' reference would be needed ... where to find it?
I am unable to suggest a quick solution on accuracy assesment issues..
Sometime I ask myself whether I am a magnet to bad data.
Or could it simply be ... am I one of the fews who does not take accuracy
for granted ?
Possibly.
After all I was the one to pick up a 1.7km error in a Quickbird scene which
many before me had overlooked and had used to vectorize a complete town
(that street-set should not be used to define accuracy).
And last time I quality assessed OSM (1 month ago) in one single screenview
I stopped counting at error #20 (not only trivial mistakes such as a
wrongly named street or a missing roads as I had expected but some serious
errors such 2 non existing steets, a minor street denoted as dual lane
arterial and an intersections where one street did not reach the
intersection by a full block. Still a long way to go.
Am I surprised? Not really. After all some of the guys with a (Garmin) GPS
have yet to realize that tracklogs they have been recording and contributing
for the past year are simply traces of their own street maps (when the
default snap-to-street function is not disabled).
Ciao
Maurizio
----- Original Message -----
From: "Marc Pfister" <[email protected]>
To: "Maurizio" <[email protected]>
Cc: <[email protected]>
Sent: Wednesday, August 12, 2009 8:13 PM
Subject: RE: [OAM-talk] Open Aerial Map Restart Meetup
Maurizio wrote:
"In my mind the main challenge of the project is data in all its aspects:
- how is data made available to the project?
- how it is screened in terms of licencing?
- who will preprocess/verify data?
- how/who will be responsible for keeping some sort of metadata records?
- how/who will tile the imagery?"
Another question is how will it be screened in terms of accuracy? I'd hate
to see some nice municipal coverage, orthorectified to a tight DEM and GCP
set, get clobbered by some UAV imagery with a 3-point warp.
In a my fantasy world, I think it would be neat to be able to connect to a
tile server and pass parameters defining accuracy, recency, resolution or
whatever, and get back matching tiles.
Marc Pfister
_______________________________________________
talk mailing list
[email protected]
http://openaerialmap.org/mailman/listinfo/talk_openaerialmap.org
