I agree in principal with what you are saying, and most of this email will boil down to semantics, but I think that what I'm saying is at least valid technically.
You say "local authorities don't buy rubbish!!" -- this is a pretty rash assumption -- I personally know instances that disprove this theory (I'm referring to non mapping related stuff). I can't comment on this instance because I have no idea about the history of this data. There are inaccuracies in any navigation system, and the demands on this stuff are pretty intense -- it's a plane flying around, computing lots of angles and velocities. there will be an error factor in there, I have no idea of it's scale, it may even be sub milimeter, but it will be there. There is also the factor with any orthorectification system of parallax, if it's not straight down imagery (and that really isn't doable, only one pixel can be in theory) there will be some parallax present. To fully correct for it, you need a full 3d map of the area, and if you already have that, there is little need of doing an areal survey. You can correct for it pretty well (using technologies such as low res 3d maps, photometry off your own images or laser scanning the area at the same time as photographing it), but it still needs to be considered. I'm sure that there are other inaccuracies that I don't even understand. Is there a detailed spec somewhere for the imagery? (I'd like to see it for personal reasons as well as for this descussion) There is also the idea cropping up here that the OS folks are beyond error, they are very good, but infallibility is going too far -- the above about there always being inaccuracies holds here also. If both surveys have been done from completely independent sources with no common error factor; compared; and found to match to a given extent, that's a pretty strong indication of accuracy, but not a guarantee. Now if we have 20 low grade GPS readings all saying a road being in one place, and the photo imagery says it's x meters to the left, I would look at the spot and ask the following questions: does it have a large metal building near it (this could consistently throw our GPS off)? is it on top of a hill or other steep slope (this could throw off orthrectification)? is there anything else that could be causing either to be off? what does the recently realised OS data say? is the error significant enough to care about?(and that one really is a question, not a statment) JR On 1 May 2010 13:04, James Rutter <jrrut...@gmail.com> wrote: > Guys, there's been some suggestions on the list that there are discrepancies > between the new air survey and previous GPS traces and also suggestions that > the image needs 'reprojecting' etc. > The image is produced by state of the art technology. As someone observed it > is made up of thousands of individual frames. They all look as if they have > odd alignments (if you look at the tiles around the collar of the image) > because this is what happens during aerotriangulation and block bundle > adjustment of air survey frames to compensate for the pitch, yaw, roll and > crabbing of the aircraft as it's flying the survey....it's not made up of > precise squares and 90 degree right angles!!. There is some extremely > expensive kit on the plane called the IMU (inertia measurement unit) and > military grade GPS (which you need a license to use!) so the air craft knows > precisely its position and attitude when each frame is shot. Independant GPS > ground control also has a bearing on the image which is 100% map accurate > and also fully orthorectified. The image is very high specification (local > authorities don't buy rubbish!!) and is in full use by Surrey County and the > districts.....any discrepancies in the accuracy and the image would not have > been accepted....and don't forget that we're checking the image accurace to > 1:1250 scale Ordnance Survey Mastermap. My polite suggestion would be that > OSM data alignment is the issue, not the postition of the air survey. > James > _______________________________________________ > Talk-GB mailing list > Talk-GB@openstreetmap.org > http://lists.openstreetmap.org/listinfo/talk-gb > > _______________________________________________ Talk-GB mailing list Talk-GB@openstreetmap.org http://lists.openstreetmap.org/listinfo/talk-gb
