[GRASSLIST:9551] Re: rectifying images

[Markus Neteler]

Phil,

it's worth trying, but probably the RAM won't suffice.
FFT images are consuming an incredible amount of memory,
the larger they are. But you could try of course.

Or loop over the image in pieces while increasing the
resolution.

Markus

On Fri, Dec 16, 2005 at 03:59:28PM -0500, Phillip J. Allen wrote:
> Markus,
> 
> Would this alogrithm be suitable to "completely"  cover two stereo-pair 
> images with all possible/likely GCP?  I am thinking of somthing to help 
> make automatic DEM's from stereo-pair images!
> 
> Phil
> 
> Markus Neteler wrote:
> 
> >hi,
> >
> >a student of us has implemented improved versions of
> >i.points and i.rectify based on previous work as published
> >in our last year's Bangkok paper.
> >
> >The new version
> >- use FFT correlation to automatically search for ground
> > control points in a user given window pair (so, just zoom
> > more or less the same zone in the master and the slave
> > image)
> >- do this for the entire stripe of images
> >- bad GCPs are identified and eliminated
> >- run the modified i.rectify to rectify the entire stripe.
> >
> >We are using this to geocode historical images. I still
> >have to cleanup the code to the current coding standards,
> >then I need a beta tester :-)
> >
> >If course you can also work with just 2 overlapping images
> >and enjoy the automated GCP search.
> >
> >Interested?
> >
> >Markus
> >
> >On Fri, Dec 16, 2005 at 03:13:03PM +0100, Wolfgang Zillig wrote:
> > 
> >
> >>I'm not sure if this helps for my problem (perhaps I was not clear 
> >>enough in my description). The problem is, that I have one original 
> >>images which I want to use as reference and all others are more or less 
> >>deformed but no one at the same level. So I think I will have mark my 
> >>reference points in all deformed images. When I understand right 
> >>grouping is good, when you have many images whith the same deformation, 
> >>or am I wrong?
> >>
> >>Wolfgang
> >>
> >>Hamish schrieb:
> >>
> >>   
> >>
> >>>>I need to rectify several images in the next time. Is there a way to
> >>>>automatize this procedure a little bit? I have the "original" image and
> >>>>then about 10 to 20 images which needs to be corrected that they refer
> >>>>to the original one. I thought of something like saving the correction
> >>>>points of the original image to an file, so that I don't need to mark
> >>>>again (the reference points will be the same for all images) and I only
> >>>>need to specify them in the images which needs to be rectified.
> >>>> 
> >>>>
> >>>>       
> >>>>
> >>>Add all the images to a group, then rectify the lot.
> >>>
> >>>i.group input=name[,name,...]
> >>>
> >>>i.rectify -a
> >>>
> >>>
> >>>
> >>>Hamish
> >>>
> >>>
> >>>
> >>>
> >>>     
> >>>
> >
> > 
> >
> 

-- 
Markus Neteler     <neteler itc it>       http://mpa.itc.it
ITC-irst -  Centro per la Ricerca Scientifica e Tecnologica
MPBA - Predictive Models for Biol. & Environ. Data Analysis
Via Sommarive, 18        -       38050 Povo (Trento), Italy