[GRASSLIST:964] Re: orthorectification

Hamish hamish_nospam at yahoo.com
Mon May 1 06:03:19 EDT 2006


> I have a set of aerian pictures which followed each others, with about
> 60% of common coverage. My goal is to obtain a unique color picture
> orthorectified and georeferenced.
> 
> 1°) I've scanned my aerian pictures to obtain tiff files. I've
> imported them in grass using r.in.gdal, so i've 3 rasters files (RGB)
> by aerian pictures. I've done the orthorectify process, picture by
> picture, working with the 3 rasters RGB of one picture at a time.
> Is it the good way to do it, or is there a faster way to do it?
> isn't it better to gather the 3 RGB rasters using r.composite and then
> going on the orthorectification process on this picture, or on a group
> of composite pictures belonging to my set of aerian pictures?

Yes, i.group will help you rectify all three images with just one set of
ground control points and is the best thing to use.

r.composite will reduce each band to 32 values from the original 256,
so you lose information when you use it. It is better to keep all
three channels and display with the d.rgb module or the ps.map equivalent.
 
> 2°) Once the orthorectify process is done, i've got many
> orthorectified and georeferenced rasters.
> So i've used r.composite to gather RGB rasters of a same picture.
> then, I've got many composite rasters, orthorectified and
> georeferenced, each raster having part of itself in common with
> another raster (common coverage of 60%) and each raster having
> uninteresting informations (fiducial marks, black strip surrounding
> the picture itself, aerian picture indicators as focal, ...) that
> needs to be suppressed. How can i do now to gather all these rasters
> to have a single accurate orthorectifed and georeferenced raster
> without the uninteresting informations, and taking advantage of the
> commun coverage of the rasters to improve the quality of this unique
> raster, for exemple in attenuating the errors due to the camera on the
> edge of each aerian picture. I've thought to use i.image.mosaic or
> r.patch, but i don't know what to do with it....
> Can i do what i wwant using Grass? or do i need another software? if
> so, do you know open source software coping with my needs

(as Maciek already has commented)
r.mapcalc to make a dynamic MASK based on row() and col() values, to
only keep the middle of each image, then r.patch? (r.patch works in
order of input maps)


> [cygwin lockups on mouse ops.]

mouse bug recently fixed.


Hamish




More information about the grass-user mailing list