Running parallel_stereo on OHRC data

[prasharajay]

ISIS version(s) affected: 10.0.0

Description

we are referring the following link
https://stereopipeline.readthedocs.io/en/latest/examples/chandrayaan2.html
As per section 8.15.1.4. Stereo we had run the parallel_stereo command with stereo-algorithm asp_bm after which we observed no stereo/run-PC.tif was generated for running further command point2dem. only file name with PC_sub.tif was present in the stereo folder

[oleg-alexandrov]

Responding a bit late as normally this should have been on the ASP issues list.

It is not clear what is going on. It is suggested to read or share the log in the terminal. Maybe it failed in interest point generation (no interest point matches), or maybe the stereo convergence angle was not good (also printed).

You can also inspect the produced L.tif and R.tif images (if it gets so far).

I heard from people at ISRO who used this path and it worked. So hoping it is a data issue or something.

If no luck, please reopen the issue on the ASP issues list.

[tckb]

@prasharajay please read the paper I wrote which talks about this process and alignment https://arxiv.org/abs/2602.14993

[oleg-alexandrov]

@tckb That is nice work. The challenges were many at this early stage of OHRC pipeline development. Aligning with a LRO NAC DEM intermediary was a good call. ASP also has the ability to do hillshade-based alignment, which can handle large shifts, if the two input DEM are gridded to same grid and look similar enough when hillshading. You did manual DEM tie points which was another good choice.

[prasharajay]

@oleg-alexandrov that issue is resolved now we were able to work thanks for your support
@tckb thanks for the sharing the paper you have nicely written the paper very good work.
i am from OHRC Data Processing team from isro.

[oleg-alexandrov]

Great. This issue can be closed then.

[tckb]

@tckb That is nice work. The challenges were many at this early stage of OHRC pipeline development. Aligning with a LRO NAC DEM intermediary was a good call. ASP also has the ability to do hillshade-based alignment, which can handle large shifts, if the two input DEM are gridded to same grid and look similar enough when hillshading. You did manual DEM tie points which was another good choice.

Thanks @oleg-alexandrov and ISIS team! I’ve been trying to make this work for many months now and had to resort to many handwritten tooling to work. The initial blog post on this at ASP was ann incredible help to write a stable pipeline that I could replicate. For alignment, I’ve tried several different approaches including LOLA. For my initial tries I did manual tie points as the automatic matching has large alignment error. Where can I reach out to you for further clarification and details ?

[tckb]

Thank you @prasharajay I have a lot questions for the OHRc team too - how do I reach out ?

[oleg-alexandrov]

"Where can I reach out to you for further clarification and details ?"

You an file an issue in ASP issues list I guess, given that is related to ASP.