Using CSM sensor models in ASP

109 views
Skip to first unread message

oscar....@digitalglobe.com

unread,
Oct 31, 2017, 11:22:38 AM10/31/17
to Ames Stereo Pipeline Support
I am trying out ASP for a project where I need point-cloud and DSM extraction from stereo and multiview imagery. However, I need to use my own sensor models derived from the Community Sensor Model (CSM). My imagery has been bundle adjusted by our own least-squares implementation, and now I want to use those adjusted rigorous sensor models in ASP. Is there some documented process for adding custom sensor models? Or perhaps there is already support for CSM models? My updated models are serialized to a file in a custom XML ISD format that is saved alongside the original geotiff (or NITF) image file. I'm new to ASP so any guidance would be appreciated. 

Alexandrov, Oleg (ARC-TI)[SGT, INC]

unread,
Oct 31, 2017, 12:18:17 PM10/31/17
to oscar....@digitalglobe.com, Ames Stereo Pipeline Support, stereo-pip...@lists.nasa.gov
Dear Oscar,

We regretfully do not support the Community Sensor Model, and there is a good chance we may not be able to add such support unless we are funded to do so. 

My best recommendation for you would be to find a way to approximate your model with an RPC model, say in the format used by WorldView. We have very good support for RPC, and other stereo packages support it as well. Since you already wrote your own least square bundle adjustment solver, writing a solver to find the best fit for some RPC coefficients seems like a doable thing to me, at least it was easy for us in ASP. If you ever consider going down that road we can have a more in-depth discussion on this subject and share our experience. 

Oleg


 

From: ames-stereo-pi...@googlegroups.com [ames-stereo-pi...@googlegroups.com] on behalf of oscar....@digitalglobe.com [oscar....@digitalglobe.com]
Sent: Tuesday, October 31, 2017 8:22 AM
To: Ames Stereo Pipeline Support
Subject: Using CSM sensor models in ASP

I am trying out ASP for a project where I need point-cloud and DSM extraction from stereo and multiview imagery. However, I need to use my own sensor models derived from the Community Sensor Model (CSM). My imagery has been bundle adjusted by our own least-squares implementation, and now I want to use those adjusted rigorous sensor models in ASP. Is there some documented process for adding custom sensor models? Or perhaps there is already support for CSM models? My updated models are serialized to a file in a custom XML ISD format that is saved alongside the original geotiff (or NITF) image file. I'm new to ASP so any guidance would be appreciated. 



This electronic communication and any attachments may contain confidential and proprietary information of DigitalGlobe, Inc. If you are not the intended recipient, or an agent or employee responsible for delivering this communication to the intended recipient, or if you have received this communication in error, please do not print, copy, retransmit, disseminate or otherwise use the information. Please indicate to the sender that you have received this communication in error, and delete the copy you received.

DigitalGlobe reserves the right to monitor any electronic communication sent or received by its employees, agents or representatives.

--
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/802d7a4f-daec-4a07-b8a1-035a16b64390%40googlegroups.com.
For more options, visit https://groups.google.com/d/optout.

Alexandrov, Oleg (ARC-TI)[SGT, INC]

unread,
Oct 31, 2017, 12:29:32 PM10/31/17
to oscar....@digitalglobe.com, Ames Stereo Pipeline Support, stereo-pip...@lists.nasa.gov
I would like to add some more to my last email. We do not have a process to add support for a new camera model, and my experience tells me it is usually not a small amount of work, and careful and extensive testing is needed. 

But, if going down that path, what is needed is a piece of C++ code that can read your model, and implementations for functions that (a) given an xyz point in planet-centric Cartesian coordinates could project it to a pixel in the camera, (b) given a pixel in in the camera can find the ray going from that pixel (in same coordinate system as before), and (c) given a pixel would find the camera center for that pixel. 

Even if you implement such functions, we may need some convincing to integrate the code and it may take us some planning to get it done. Ideally it would be some generic standard that may get wider use. And I'd still think your best bet is an RPC approximation, those tend to work very well, and they can also done piecewise, if it is felt that the image is too big for a single RPC model to be good for the entire extent, and you'd have total control over it.

Oleg



From: ames-stereo-pi...@googlegroups.com [ames-stereo-pi...@googlegroups.com] on behalf of Alexandrov, Oleg (ARC-TI)[SGT, INC] [oleg.al...@nasa.gov]
Sent: Tuesday, October 31, 2017 9:18 AM
To: oscar....@digitalglobe.com; Ames Stereo Pipeline Support
Cc: stereo-pip...@lists.nasa.gov
Subject: RE: Using CSM sensor models in ASP

Reply all
Reply to author
Forward
0 new messages