Warping stuck at 0

[Shuwen-Liu]

AROSICS version: 1.12.1
Python version: 3.10.18
OS: Windows 11

Dear arosics team,

First of all thank you so much for creating this package which is really helphul!

I encounted some issues with several of my images. Specically, warping will stuck at 0 forever when I use CRL.correct_shifts() or DeShifter.DESHIFTER(), but do not meet problem with function do not have warping step, e.g. CRL.calculate_spatial_shifts() or CRL.view_CoRegPoints().

I have no issues with my other images so I didn't figure out where is the problem.

I attached the output of the script, and the reference/target image in this google drive folder. https://drive.google.com/drive/folders/11c8JTDbxGnNuGLsnRw5YkF_YzW1vn5T5?usp=drive_link

I really appreciate if someone can help me with this issue. Thanks in advance!

Best,

Shuwen

[danschef]

My guess is that it hangs when running the GDAL code under the hood of AROSICS which is used for image warping. Might be due to limited memory resources on your machine and could also be fixed with a recent GDAL version. Your Python version seems a bit outdated, so you try to install AROSICS on a recent Python version and the latest version of GDAL (should be installed automatically when creating a new conda environment from scratch).

Apart from that, you could also have a look at your free memory when running the image warping. Using less tie points would reduce the memory load.

However, without knowing your actual call of AROSICS and the corresponding output, I can only guess.

[Shuwen-Liu]

Hi @danschef , really appreciate your help. I have tried install Python 3.13 and GDAL 3.11 but it didn't work. Memory do not have issue either.

This is the setting I used. The reference and target image could be found in the google drive folder.

kwargs = {
'grid_res': 30,
'window_size': (256, 256),
'nodata': (-9999, -9999),
'path_out': 'auto',
'max_shift': 100,
'fmt_out': 'GTiff',
}

reference_image_fp = xxx
target_image_fp = xxx
CRL = COREG_LOCAL(reference_image_fp, target_image_fp, **kwargs)

[danschef]

Ok, I can try to reproduce it after my vacation, end of July.

[Shuwen-Liu]

Hi @danschef, truely thanks for that!
I now use a workaround to fix this issue by exporting GCPs from coreg_info and using gdal.Warp() to do the warping, instead of calling CRL.correct_shifts() or DeShifter.DESHIFTER().

However I notice some difference bewteen the warping results of arosics and gdal.warp, when I test it on my scusessful images. I set 3rd order polynominal transformation when used gdal.warp. Arosics document said it use polynominal transformation without detailed information. Does other information in coreg_info other than GCPs cause the difference? Or the transformation method? And do you have strong recommendations to use arosics warp?

Thanks again for taking time to response during your vacation.