[gdal-dev] pansharpen subset of multispectral bands

[Even Rouault]

Le 09/01/2025 à 20:27, Joe McGlinchy via gdal-dev a écrit :
> Hello,
>
> I'm wanting to pansharpen only a subset of multispectral bands I have 
> available. Say for instance I have blue-green-red-nir, and a 
> panchromatic band, I can calculate the relative weights to specify 
> when using `gdal_pansharpen` on the full multispectral dataset. If I 
> want to only pansharpen, for example, the red and NIR bands, is it 
> best to perform the pansharpening on the full dataset and only retain 
> the bands I need? Or can i provide only the weights specific to the 
> bands I want to pansharpen? If I can do the latter (which I can, given 
> the parameter set to `gdal_pansharpen`), do I need to recalculate 
> those weights, or only provide the subset of weights for those 
> specific bands?

Given the formula at 
https://gdal.org/en/stable/drivers/raster/vrt.html#gdal-vrttut-pansharpen

pseudo_panchro[pixel]  =  sum(weight[i]  *  spectral[pixel][i]  for  i=0  to  nb_spectral_bands-1)
ratio  =  panchro[pixel]  /  pseudo_panchro[pixel]
for  i=0  to  nb_spectral_bands-1:
     output_value[pixel][i]  =  input_value[pixel][i]  *  ratio


if you only provide a subset of the input spectral bands, you won't be 
able to reconstruct a plausible pseudo_panchro values, hence the scaling 
ratio will not be accurate.


If your spectral dataset has bands in the R,G,B,NIR order, you can 
achieve what you want with:


$ gdal_pansharpen panchro.tif rgbnir.tif,band=1 rgbnir.tif,band=2 
rgbnir.tif,band=3 rgbnir.tif,band=4 out.tif -b 1 -b 4

Even

-- 
http://www.spatialys.com
My software is free, but my time generally not.
Butcher of all kinds of standards, open or closed formats. At the end, this is just about bytes.
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.osgeo.org/pipermail/gdal-dev/attachments/20250109/74a6aa33/attachment.htm>