YES
My mistake was thinking that the map projection was the source map, not the
target. Ofcourse the target needs to be in WGS84 when the track and
waypoints are...
So that fixed it and the tracks now seem to be within GPS accuracy.
Thank you!
/Henrik
On 16 March 2011 13:22, Oliver Eichler wrote:
> Hi Henrik,
>
> a track should be on the road as on any other map. If all points are offset
> by a certain value, you very likely made a mistake with the map datum.
>
> Here are just my 2 cents:
>
> * 1200dpi is way too much. Imho 600dpi is way enough. Keep in mind that
> elements on the map are generalized. Thus a street has a certain wigth
> because it's a street of a certain kind. That has nothing to do with the
> real width. Also you will net get more accuracy of points in the map by
> up-sampling the map.
>
> * the line with the map projection is the target projection. That is not
> necessarily the map's original projection. I would take something easy for
> the beginning like "+proj=merc +datum=WGS84".
>
> * the ref. projection is the projection and datum of your reference points.
> Some maps have a grid that can be taken as reference. In that case you have
> to supply the grids projection and datum. However, as you use road
> crossings, I would use "+proj=longlat +datum=WGS84" and take the reference
> points from a vector/osm map loaded in QLGT via clipboard.
>
> * using crossings, especially road/railroad, as reference points is ok. Not
> perfect but in most cases sufficient. But select the points as much apart
> from each other as possible.
>
> * thin plate is not what you want. It's a very special case. Use linear or
> quadratic.
>
> * European datum. Many maps use the one from 1950. But that shouldn't cause
> 120m offset.
>
> HTH
>
> Oliver
>
> -------- Original-Nachricht --------
> > Datum: Wed, 16 Mar 2011 04:31:48 -0700
> > Von: qlndkartegt-users.hkristen...@xoxy.net
> > An: qlandkartegt-users@lists.sourceforge.net
> > Betreff: [Qlandkartegt-users] GeoTIFF calibration accuracy - Now in plain
> text...
>
> >
> > What kind of calibration accuracy should I expect from a scanned map?
> >
> > I'm not totally happy with the results I'm getting - but I have only been
> > a GPS owner for 24h so it is quite likely than I'm not doing the right
> > things...
> >
> > This my issue:
> >
> > I scanned an official Danish topographical map.
> > Scale 1:25000
> > Datum: "European" (no year stated but copyright 1996, so I'm asuming
> 1987)
> > Projection and grid: UTM zone 33
> >
> > So I go to Map Edit/Create, Convert Tiff to GeoTiff
> > -enter the scanned source file (which as a PNG, not TIFF)
> > -set Map Projection to UTM 33 and European 1987 datum
> > -set Reference Projection to UTM 33 and European 1987 datum
> > -set mode to Thin Plate (4 ref pts)
> > -enter reference point coordinates as UTM easting and northing from the
> > map grid
> >
> > As reference points I use 4 grid intersections on the map forming a
> > rectangle of 4x3km.
> > I have scanned at 1200 dpi and when zooming in it looks like I can set
> the
> > reference points on the map with an accuracy of app. 2 meters.
> >
> > After the conversion finishes I then load the map.
> >
> > I then load a track and some waypoints from my eTrex (downloaded using
> > gebabbel).
> > (the eTrex was set to European 1979 datum while recording, but as I
> > understand it that shouldn't matter as it always converts and exports as
> WGS84 ?)
> >
> > The track and waypoints show up nicely on the map - but off by about 120
> > meters...
> >
> > Is that as good as it gets?
> > -Or am I doing something wrone?
> >
> > PS: I have tried using European 1979 datum instead. And 6 reference
> > points. But that doesn't make any noticeable difference.
> >
> > TIA,
> > Henrik
> >
>
------------------------------------------------------------------------------
Colocation vs. Managed Hosting
A question and answer guide to determining the best fit
for your organization - today and in the future.
http://p.sf.net/sfu/internap-sfd2d
_______________________________________________
Qlandkartegt-users mailing list
Qlandkartegt-users@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/qlandkartegt-users
