Orthorectifying GeoEye Data

[TimStahl]

Hi all,

I'd like to orthorectify a pansharpened or composite band raster from a GeoEye dataset.  The RPC files allow easy rectification of each individual band, but it appears that these do not carry over to the new rasters.  The RPC files are .txt with format "rastername_rpc.txt"  and I tried doing the same for the pansharpened/composite with no success.  Do I need to orthorectify each band and then create the composite rasters?

Cheers,
Tim

[JeffreySwain]

I would recommend following the instructions here, which refer to either using the Create Ortho-corrected Raster Dataset, geometric function in a mosaic dataset or use the ortho rectify button on the Image Analysis Window.

So you should not have to do each band individually, but you will have to follow one of the three paths.  Are you experiencing the bad solution after following one of them?

[ChrisM]

This is a few years after the original post, but I am having a similar issue. The directions at that link do not provide a description of how to associate the composites (or even more complicated, a mosaic from the composites) with the individual rpc files for each coverage. Doing each band with the rpc file and the dem is easy but not the composites (or mosaics). Suggestions? Am I SOL on the mosaic but just need to copy the rpc files over to to the folder with the composite data?

Cheers!

Chris

[NeilAyres]

I did this a while ago.
After much reading and messing about I ended up with a workflow like this :
1. Orthorectify each band.
2. Create a 4 band composite.
3. Do the pan-sharpening.

See the pic as an illustration.

Good luck,
Neil

[ChrisM]

Thanks! I suppose that is what I was trying to avoid: doing each band. I ended up just orthorectifying the panchromatic images.

[JeffreySwain]

So you could not use the Image Analysis window, mosaic dataset or tool to process this?  The workflow expects multiple bands and having the RPCs per band should not be a problem.  I would try the image analysis window or mosaic dataset. You probably should not have had to process each band and if you did, then I think that Support would like to see the data to pass it to development. 

Mosaic Dataset

Image Analysis Window

The software should be able to recognized the scene normally and then apply the geometric function to the whole scene.

[NeilAyres]

Jeff,

as you can see from the function chain pic attached, I was using the image analysis window.

As I see it, because you get an "rpc" file per band, you have to do the ortho rectify on each band first then pass that on the composite / pansharp function afterwards.

I have found that doing everything, all together in a mosaic is not always (well, lets be honest, never) the best way.

Generally I do all the messing about using the image analysis workflow. Then export out the final version into a fgdb, then mosaic the various scenes.

N

[JeffreySwain]

I see that pic now, I apologize, I am still sorting out the GeoNet-ness and missed it initially.  I do not think you should have to do it that way.  I think the design is to use a chain like Composite Bands > Geometric > Pansharpened (it would tier in the chain so you would only have to do it once rather than for each band) and only use it once.  I would think that Development would like to know if you are having to jump through such hoops.  After all this is just one scene, the if you applied the functions in a mosaic dataset, then this might be problematic unless you applied the chain during the import of the scenes to the mosaic dataset.  To get to Development, a call to Support might be in order.  However if you are fine with your workflow, then I suppose it is not necessary.  I see this as a potential bug in the software, not simply a workaround given your data. 

Thanks for sharing.