internal corners in checkerboard pattern

26 views
Skip to first unread message

Daniel Falaschi FFyL UNCuyo

unread,
Apr 23, 2024, 11:02:33 AMApr 23
to Ames Stereo Pipeline Support
Hello all, I am new in the use of Ames Stereo Pipeline, and I am trying to figure out a reasonable workflow for the processing of historical aerial photos with limited calibration information. As a first step, I am using the camera_calibrate tool. Among the options to be set are the so called ¨number of internal corners in checkerboard pattern¨. Does tis mean the fiducial marks? If so, should I set the board height and width to 2 and 2?

Many thanks for your help.

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

unread,
Apr 23, 2024, 11:48:08 AMApr 23
to Daniel Falaschi FFyL UNCuyo, Ames Stereo Pipeline Support
The camera_calibrate tool assumes you have access to the camera and can actually calibrate it using a checkerboard (https://docs.opencv.org/4.x/dc/dbb/tutorial_py_calibration.html).

Likely you don't have access to the cameras. Fiducial marks cannot be used for that purpose.

Handling historical images can be a pain. Our introductory document is at https://stereopipeline.readthedocs.io/en/latest/sfm.html#sfm-examples-with-orbital-images. We have specialized pages at https://stereopipeline.readthedocs.io/en/latest/examples.html for declassified images.

I will suggest in particular great caution if you plan to process the panoramic (optical bar) data such as from KH satellites as our process for that is not fully robust. 



From: ames-stereo-pi...@googlegroups.com <ames-stereo-pi...@googlegroups.com> on behalf of Daniel Falaschi FFyL UNCuyo <danielf...@ffyl.uncu.edu.ar>
Sent: Tuesday, April 23, 2024 8:02 AM
To: Ames Stereo Pipeline Support <ames-stereo-pi...@googlegroups.com>
Subject: [EXTERNAL] [BULK] internal corners in checkerboard pattern
 
CAUTION: This email originated from outside of NASA.  Please take care when clicking links or opening attachments.  Use the "Report Message" button to report suspicious messages to the NASA SOC.



Hello all, I am new in the use of Ames Stereo Pipeline, and I am trying to figure out a reasonable workflow for the processing of historical aerial photos with limited calibration information. As a first step, I am using the camera_calibrate tool. Among the options to be set are the so called ¨number of internal corners in checkerboard pattern¨. Does tis mean the fiducial marks? If so, should I set the board height and width to 2 and 2?

Many thanks for your help.

--
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/63e78d97-63b2-430c-bd94-8528b526ab13n%40googlegroups.com.
Message has been deleted

Daniel Falaschi FFyL UNCuyo

unread,
Jun 21, 2024, 11:51:17 AM (12 days ago) Jun 21
to Ames Stereo Pipeline Support
Many thanks Oleg for your reply. I am back to AMS but it is a real struggle.
I´ll try to better illustrate the kind of processing that I have in mind.
I want to generate DEMs and orthos from historical aerial photographs. To do this, I only have the focal length and distances between the fiducial marks in mm.
I understand that the first step in the processing would be the solving of the camera intrinsics by  generating an initial camera tsai file, using trivial values for the camera center and rotation matrix.
The tsai file is generated using the cam_gen tool. Are the longitude-latitude coordinates of the camera image corners compulsory? I mean, If I must process 20 individual images, do I need corner coordinates for all of them? Or can I use a GCP file instead for this?
In the current asp_book pdf (section 16.8), I cannot see how I can pass ALL the parameters, for example the distortion model (NULL in my case) and the rotation matrix.  

Many thanks for your assistance

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

unread,
Jun 21, 2024, 12:39:35 PM (12 days ago) Jun 21
to Daniel Falaschi FFyL UNCuyo, Ames Stereo Pipeline Support
>The tsai file is generated using the cam_gen tool. Are the longitude-latitude coordinates of the camera image corners compulsory? I mean, If I must process 20 individual images, do I need corner coordinates for all of them? Or can I use a GCP file instead for this?


An alternative is not to use cam_gen, but create a solution using SfM, if all your images have good overlap, and use a handful of GCP only later, to register to the ground.  https://stereopipeline.readthedocs.io/en/latest/sfm.html#sfm-examples-with-orbital-images

> In the current asp_book pdf (section 16.8), I cannot see how I can pass ALL the parameters, for example the distortion model (NULL in my case) and the rotation matrix.  

The rotation matrix will be determined for you by cam_gen based on lon-lat corners. NULL distortion will be created for you as well. The distortion can be added to the produced files manually, once you validated that the cameras are decently positioned. https://stereopipeline.readthedocs.io/en/latest/pinholemodels.html#pinholemodels



Sent: Tuesday, June 18, 2024 9:44 AM

To: Ames Stereo Pipeline Support <ames-stereo-pi...@googlegroups.com>
Subject: Re: [EXTERNAL] [BULK] internal corners in checkerboard pattern
 
Reply all
Reply to author
Forward
0 new messages