[gdal-dev] Combine aerial images

[Jukka Rahkonen]

Frank Broniewski <brfr <at> metrico.lu> writes:

> 
> Hi,
> 
> I have a directory with appr. 2800 aerial images, each 1kmx1km. I'm 
> using this source for a Mapserver WMS layer. Unfortunately the amount of 
> data to process is way too much for a Mapserver WMS service.
> 
> Now I want to create larger overview images where the aerial images are 
> combined together. The route I wanted to was
> 
> 1) build a VRT from the images
> 2) use gdal_translate to get a new geotiff with a smaller size (pixel wise)
> 
> Is this a valid route to go?

Yes it is.

I have configured Mapserver orthoimage layers so that Mapserver does not 
ever need to to open more that about 12-16 original tiff images per one 
screenful of output. We have 12000x12000 pixel images with 0.5 m pixel 
size. Four such side by side makes 48000 pixels and if they are pressed
to 1000 pixel wide output image the pixel size in output is roughly 25 m.
Therefore I have prepared resampled layer with 25 m pixel size which 
is used for small scale output.

Prepare new set of images, each covering 50x50 kilometers or more 
with 50 times bigger pixels than the originals. Create scale 
dependent GROUP layer for Mapserver and you should get good speed.
Naturally you must run gdaladdo for the resampled images and perhaps
make one more re-resampled set if your coverage is very big. 

-Jukka Rahkonen-

> 
> Many thanks,
> 
> Frank