BA on hirise

20 views
Skip to first unread message

Mauro S

unread,
Jul 15, 2022, 10:34:25 AM7/15/22
to Ames Stereo Pipeline Support

Hello again,

I'm trying to run BA with two HIRISEs and the result seems quite strange, the point recognized are not well clearly. Iḿ runnining the BA as

bundle_adjust A.cub B.cub -o ba_HIRISE/runBA

Being A and B the outputs from hiedr2mosaic.py

Do you think that playing with BA parameters can solve it?

Screenshot from 2022-07-15 08:29:35.png

Alexandrov, Oleg (ARC-TI)[KBR Wyle Services, LLC]

unread,
Jul 15, 2022, 12:12:55 PM7/15/22
to Mauro S, Ames Stereo Pipeline Support
I am not sure what you are plotting here. Normally you should run:

stereo_gui A.cub B.cub ba_HIRISE/runBA-A__B.match

to see the matches, given your choice of notation. This should work. This functionality has been around for a while and used frequently. 
From: ames-stereo-pi...@googlegroups.com <ames-stereo-pi...@googlegroups.com> on behalf of Mauro S <mgs...@gmail.com>
Sent: Friday, July 15, 2022 7:34 AM
To: Ames Stereo Pipeline Support <ames-stereo-pi...@googlegroups.com>
Subject: [EXTERNAL] BA on hirise
 
--
You received this message because you are subscribed to the Google Groups "Ames Stereo Pipeline Support" group.
To unsubscribe from this group and stop receiving emails from it, send an email to ames-stereo-pipeline...@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/ames-stereo-pipeline-support/cb7c7bff-9632-401d-bcb2-7ae8d4c72e5an%40googlegroups.com.

Mauro S

unread,
Jul 15, 2022, 2:08:39 PM7/15/22
to Ames Stereo Pipeline Support
Hello Oleg,

yes, sorry I did not mentioned it  but the image I showed is exactly the view from the line you suggested.

I think something wrong went in the Bundle adjustment because there are no good matches

I run again the BA and now I see that this errorappears at the end:
Writing: ba_HIRISE/runBA-ESP_012438_1795_RED.mos_hijitreged.norm__ESP_012728_1795_RED.mos_hijitreged.norm-clean.match
Error: Too few points remain after filtering!.

Alexandrov, Oleg (ARC-TI)[KBR Wyle Services, LLC]

unread,
Jul 15, 2022, 2:19:47 PM7/15/22
to Mauro S, Ames Stereo Pipeline Support
I am not sure what is going on. 

Are you having this error with other HiRISE datasets as well? 

It appears that bundle_adjust is really having a hard time with your images. 

You can try to experiment a bit. For example, one can first use the ISIS crop command to cut out two clips which are quite similar from both images, getting new .cub files for just those clips.

Then, one can do cam2map on those clips, and overlay the resulting georeferenced cub files in stereo_gui (there's a menu option for viewing overlaid georeferenced images). This would check if the camera orientations are at least correct. Ideally those two cam2map-ed clips would be very similar.

Then, you can try bundle_adjust on just the two clips (before cam2map). 

If no luck so far, that would be quite bad. if that works, you can try going to the original images and do 

bundle_adjust --ip-per-image 50000 left.cub right.cub -o ba/run

I chose here an outrageous number of interest points, which may be slow, but hopefully some will be left after filtering. 

You can also specify with bundle_adjust above --ip-inlier-factor 0.5, which will make it less likely to throw out matches.

Let me know if anything works or not.




Sent: Friday, July 15, 2022 11:08 AM

To: Ames Stereo Pipeline Support <ames-stereo-pi...@googlegroups.com>
Subject: Re: [EXTERNAL] BA on hirise
 
Reply all
Reply to author
Forward
0 new messages