[GRASS-user] pansharp fusion of ikonos image

Hi,

have you any suggestion on how can i do a ikonos "pansharp fusion" using grass ?
there isn't a command that already done this things,
is possible to use an r.mapcalc expression to do it ? have you experience about ?
thanks for any help!
Massimo.

On Mon, 2007-02-26 at 19:21 +0100, Massimo Di Stefano wrote:

Hi,

have you any suggestion on how can i do a ikonos "pansharp fusion"
using grass ?
there isn't a command that already done this things,
is possible to use an r.mapcalc expression to do it ? have you
experience about ?
thanks for any help!
Massimo.

A technique I've used in the past is to use i.rgb.his to convert to HIS
colorspace and replace the intensity (I) channel with the pan imagery.
Then use i.his.rgb to convert back to RGB colorspace.

--
Brad Douglas <rez touchofmadness com> KB8UYR/6
Address: 37.493,-121.924 / WGS84 National Map Corps #TNMC-3785

Hi Massimo,

please have a look at
http://grass.itc.it/grass63/manuals/html63_user/imageryintro.html
-> Image fusion

or
g.manual imageryintro
-> Image fusion

I agree that it is minimalistic, please suggest how to improve the text
(text snippets welcome!).

So, yes, you have at least two possibilites besides r.mapcalc.

Markus

Massimo Di Stefano wrote:

Hi,

have you any suggestion on how can i do a ikonos "pansharp fusion"
using grass ?
there isn't a command that already done this things,
is possible to use an r.mapcalc expression to do it ? have you
experience about ?
thanks for any help!
Massimo.

_______________________________________________
grassuser mailing list
grassuser@grass.itc.it
http://grass.itc.it/mailman/listinfo/grassuser

--
View this message in context: http://www.nabble.com/pansharp-fusion-of-ikonos-image-tf3294899.html#a9168761
Sent from the Grass - Users mailing list archive at Nabble.com.

Just as a heads up, most pansharpening algorithms introduce false spectral
information into the mix -- you might want to look into principle components
pansharpening algorithms, which tend to preserve the spectral values better
than the other commonly used ones...

http://scholar.google.com/scholar?hl=en&lr=&q=welch+merging+spot&btnG=Search

The first hit is the paper that describes this (Welch and Ehlers, 1987)

The basic concept is you rotate a resampled "high resolution" multispectral
data (I think a nearest neighbor resample to match the high res resolution
is sufficient) into PCA space, the first PCA axis tends to be the albedo
band (which is the assumption for this analysis). The high resolution pan
band is then substituted in to the PCA stack, replacing the first PCA axis
band, and then rotated back into spectral space.

I used this pansharpened data as the basis for several published papers
using IKONOS imagery, and it works very nicely (I used ENVI's implementation
of the algorithm)

--j

On 2/26/07 12:48 PM, "Markus Neteler" <neteler@itc.it> wrote:

Hi Massimo,

please have a look at
http://grass.itc.it/grass63/manuals/html63_user/imageryintro.html
-> Image fusion

or
g.manual imageryintro
-> Image fusion

I agree that it is minimalistic, please suggest how to improve the text
(text snippets welcome!).

So, yes, you have at least two possibilites besides r.mapcalc.

Markus

Massimo Di Stefano wrote:

Hi,

have you any suggestion on how can i do a ikonos "pansharp fusion"
using grass ?
there isn't a command that already done this things,
is possible to use an r.mapcalc expression to do it ? have you
experience about ?
thanks for any help!
Massimo.

_______________________________________________
grassuser mailing list
grassuser@grass.itc.it
http://grass.itc.it/mailman/listinfo/grassuser

--
Jonathan A. Greenberg, PhD
Postdoctoral Scholar
Center for Spatial Technologies and Remote Sensing (CSTARS)
University of California, Davis
One Shields Avenue
The Barn, Room 250N
Davis, CA 95616
Cell: 415-794-5043
AIM: jgrn307
MSN: jgrn307@hotmail.com

Massimo Di Stefano wrote:

Hi,

have you any suggestion on how can i do a ikonos "pansharp fusion" using
grass ?

Besides the i.fusion.brovey in GRASS, which Markus mentioned, you might
like to have a look at an OpenEV addon fusion.py from
http://openev.sourceforge.net/mariotools/. It supports Ikonos, among
the others. I liked it performance, but I'm absolutely no expert here.

Maciek

Interesting method. I've never tried that one, but it makes complete
sense. I may have to write a module to "rewind" i.pca (added to bottom
of nearly infinite queue :-).

Another method of pan-sharpening:
- Run RGB through i.rgb.his
- Replace the intensity channel with the pan image
- Change resolution to the pan image
- Revert to RGB with i.his.rgb.

This method generally preserves spectral characteristics.

On Mon, 2007-02-26 at 13:02 -0800, Jonathan Greenberg wrote:

Just as a heads up, most pansharpening algorithms introduce false spectral
information into the mix -- you might want to look into principle components
pansharpening algorithms, which tend to preserve the spectral values better
than the other commonly used ones...

http://scholar.google.com/scholar?hl=en&lr=&q=welch+merging+spot&btnG=Search

The first hit is the paper that describes this (Welch and Ehlers, 1987)

The basic concept is you rotate a resampled "high resolution" multispectral
data (I think a nearest neighbor resample to match the high res resolution
is sufficient) into PCA space, the first PCA axis tends to be the albedo
band (which is the assumption for this analysis). The high resolution pan
band is then substituted in to the PCA stack, replacing the first PCA axis
band, and then rotated back into spectral space.

I used this pansharpened data as the basis for several published papers
using IKONOS imagery, and it works very nicely (I used ENVI's implementation
of the algorithm)

--j

On 2/26/07 12:48 PM, "Markus Neteler" <neteler@itc.it> wrote:

>
> Hi Massimo,
>
> please have a look at
> http://grass.itc.it/grass63/manuals/html63_user/imageryintro.html
> -> Image fusion
>
> or
> g.manual imageryintro
> -> Image fusion
>
> I agree that it is minimalistic, please suggest how to improve the text
> (text snippets welcome!).
>
> So, yes, you have at least two possibilites besides r.mapcalc.
>
> Markus
>
>
> Massimo Di Stefano wrote:
>>
>> Hi,
>>
>> have you any suggestion on how can i do a ikonos "pansharp fusion"
>> using grass ?
>> there isn't a command that already done this things,
>> is possible to use an r.mapcalc expression to do it ? have you
>> experience about ?
>> thanks for any help!
>> Massimo.

--
Brad Douglas <rez touchofmadness com> KB8UYR/6
Address: 37.493,-121.924 / WGS84 National Map Corps #TNMC-3785