[Xastir] Re: Starting with a scanned USGS 7.5" paper map....

[Tom Russo]

On Sat, Jan 19, 2008 at 10:49:21PM +0000, we recorded a bogon-computron collision of the <rez at touchofmadness.com> flavor, containing:
> On Sat, 2008-01-19 at 11:27 -0700, Tom Russo wrote:
> > On Sat, Jan 19, 2008 at 12:12:22PM -0600, we recorded a bogon-computron collision of the <RiverRidge at CenturyTel.net> flavor, containing:
> > > I am collecting the USGS topographic maps for  my area and I am fortunate 
> > > in that all except one were on Libre Map.  The one that was missing, The 
> > > good folks at UW Madison Geography Library had the paper map and scanned it 
> > > for me.  So now I have a 454MB scanned image in tif format.
> > 
> > Nice.
> > 
> > > Setting aside the fact that for one map, I would probably be better off 
> > > just buying the proper Geotiff map files............  How do I go from this 
> > > scanned image to a georeferenced digital file?  Can I get there from here?
> > 
> > Yes, but you need a tool that you probably don't have yet.  There are two
> > tools of choice, GRASS (http://grass.itc.it/) and QGIS (http://www.qgis.org/).  
> > GRASS can do more with your data, but QGIS has a simpler georeferencing tool.  
> > GRASS has a learning curve as steep that looks a lot like Everest, QGIS is
> > a bit of a PITA to install but is comparatively easy to use.
> 
> +1.  My opinion is a bit partisan, being a GRASS developer. ;-)

Heh.

> GDAL should be all you need to convert file formats and do basic
> warping.  It is very comprehensive, but lacks any GUI for GCP selection.

Agreed, as I've been using GRASS for years and am familiar with it enough
that I use it for everything.  But it is not the tool I would recommend to a 
person who wanted nothing more than to georectify a single image.  

GRASS *should* be all one needs if one is already familiar with it, but for
a one-shot effort such as this, something like QGIS could be easier.

The point is sorta moot here since Jim is in the process of uploading the image
to someplace where I can get at it, and I will be doing the georectifying in
GRASS.

> Learning GRASS is a fairly involved process.  It is generally geared
> toward research as opposed to ease of use.  I new wxPython GUI is
> actively been developed and should be mature by GRASS v7.

So far, I've found that all the attempts at putting a GUI on grass have gotten
in my way, but gis.m is getting pretty good.  I haven't touched wxPython yet,
because I still prefer to just use command line tools.

It's taken me a long time using GRASS casually to be able to use it for the
things I need to use on the very infrequent occasions when I need it.  But
it's definitely my tool of choice.

> One of these days, I really need to put together some tutorials geared
> towards Xastir.  I'm usually too busy on other projects. :(

I've often posted here about how to do this very operation, but at this point
I've concluded that GRASS is a great tool that I'll swear by, but not ever
again recommend for casual use to someone who just wants to get one thing done.

> [snip]
> 
> > Yes, it does.  Georeferencing scanned images is tricky and time consuming, so
> > it is expected that few will want to do it.  I've done it many times, and I
> > try to avoid doing it if possible.  But with the right tools and a little
> > care it can be done.
> 
> This largely depends on your image size (cell count), resolution and
> GCPs.  It can take anywhere from a few seconds to hours.

The time-consuming part is doing all the steps to get the thing set up.  For
example, if one were to choose the 16 lat/lon graticule marks in a USGS
DRG, one would have to convert each of the 16 points to UTM, because
the desired end product is a raster georeferenced in the coordinates used for
the projection --- using the lat/long values directly and rectifying into a
lat/lon location would give bad results (and is usually the first thing someone
tries to do).  So the coordinates have to be read off the image, converted,
and then one must painstakingly click on the exact points in the image where
the graticule lines cross, then enter the data.  It's not hard, nor is it
exactly rocket surgery, but it's time consuming and there is a lot of room
for error that will result in a poorly rectified image.  Naturally, one can
short-cut the process and choose fewer GCPs for this particular task.

-- 
Tom Russo    KM5VY   SAR502   DM64ux          http://www.swcp.com/~russo/
Tijeras, NM  QRPL#1592 K2#398  SOC#236 AHTB#1 http://kevan.org/brain.cgi?DDTNM
"And, isn't sanity really just a one-trick pony anyway? I mean all you get is
 one trick, rational thinking, but when you're good and crazy, oooh, oooh,
 oooh, the sky is the limit!"  --- The Tick