Can I use alos/prism with ASP?

[asuka t]

I am a student who has just started my research in Japan and do not have much knowledge about ASP. Can JAXA's ALOS/PRISM be used with ASP? I apologize if I missed that point in the manual, or if it is too obvious a question.

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

Apparently PRISM stands for "Panchromatic Remote-sensing Instrument for Stereo Mapping ". So in principle ASP should work for it.

In practice, I don't know. You'd have to study its documentation. In ASP we expect the camera information to be either in RPC format, like for the European Pleiades, for example, or to have the camera matrix and intrinsics in a text file, like for pinhole cameras. (Our doc has more info.)

If PRISM data can be represented in a format ASP can understand, you can use it. If not, work will be needed. We could do such work only if we are funded for it.

Oleg

---

From: ames-stereo-pi...@googlegroups.com <ames-stereo-pi...@googlegroups.com> on behalf of asuka t <aska19...@gmail.com>
Sent: Friday, May 14, 2021 11:14 AM
To: Ames Stereo Pipeline Support <ames-stereo-pi...@googlegroups.com>
Subject: [EXTERNAL] Can I use alos/prism with ASP?

 

I am a student who has just started my research in Japan and do not have much knowledge about ASP. Can JAXA's ALOS/PRISM be used with ASP? I apologize if I missed that point in the manual, or if it is too obvious a question.

--
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/c2c36f90-2ac4-4dae-8fa8-dd85885c25a4n%40googlegroups.com.

[Shashank Bhushan]

Hi Asuka,

You can use the RPC model tutorial in ASP (https://stereopipeline.readthedocs.io/en/latest/examples.html?highlight=rpc#rpc-images-including-geoeye-airbus-cartosat-1-and-perusat-1) for processing ALOS PRISM stereo imagery. I used sample PRISM data very long ago (not with ASP though), and I think they are accompanied with RPC camera models which ASP can use :)

Best,

Shashank

On Fri, May 14, 2021 at 11:17 AM asuka t <aska19...@gmail.com> wrote:

I am a student who has just started my research in Japan and do not have much knowledge about ASP. Can JAXA's ALOS/PRISM be used with ASP? I apologize if I missed that point in the manual, or if it is too obvious a question.

--

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/c2c36f90-2ac4-4dae-8fa8-dd85885c25a4n%40googlegroups.com.

--

Shashank Bhushan

Graduate Student/Research Assistant

Civil & Environmental Engineering

University of Washington, Seattle

[Jordan Brown]

Hello everyone,

I've been working with this data as well recently, in the distribution from the European Space Agency, which does not include RPCs. (See p. 12 of the PDF linked here.) But it seems that some physical camera parameters are available scattered around the web, as well as some limited map projection information in the CEOS files available through the ESA themselves.

No free lunch, it would seem, but I'll keep poking around and see what I can do treating the images as though they were similar to the historical imagery cases (and perhaps SkySat) discussed in your excellent documentation!

Wishing you all the best and many thanks for your expertise and enthusiasm,

Jordan

[Oleg Alexandrov]

Give it a try. It could work. At the risk of repeating the documentation, perhaps, the biggest unknown is the focal length. Hopefully you could find some info about it, or pull one value out of the hat to start with. The optical center can be set to half the image dimensions. Once you create a simple tsai file wih the intrinsics (optical center, focal length, and the distortion, which can be set to zero), and if you have a reference DEM, such as SRTM, one can pick GCP using stereo_gui and use bundle adjustment create rough cameras, which can be verified with mapproject (or one can use camera_solve).

If you are able to get a DEM using ASP and these cameras, and it looks that the height of mountains is either too small or too large that is a symptom of the focal length being wrong. You can either try adjusting it or use bundle_adjust with --reference-dem and --reference-dem-weight to solve for it (there is some info in the doc).

Lastly, with SkySat we had a lot of issues because this camera has a very small ground footprint (~2km) and a very large focal length, which makes the camera pose determination quite fragile and would often fail to be even in the ballpark. Things worked better for KH9 whose ground footprint was, if I recall right, at least a few tens of km.

[Shashank Bhushan]

Hi Jordan, 

As Oleg suggests, this can work, you would need to get information on focal length and pixel pitch which would simplify things a lot. Focal length seems to be around 2 m and the pixel pitch is around 5.5 micron to 7 micron for along-track and cross-track viewing direction (https://www.isprs.org/proceedings/xxxiii/congress/part4/779_XXXIII-part4.pdf). I do not fully appreciate these two different values of pixel pitch in the along-track/cross-track direction, but my guess is that it has to do with a different number of CCDs filling up the focal length in the x and y direction. Maybe Oleg can provide a better handle of what value to use. 

1 thing to keep in mind is that PRISM is a linescanner (https://www.eorc.jaxa.jp/ALOS/en/about/prism.htm) so the frame camera approximation as SkySat and historicals will be tricky. However, there have been attempts in the past which have got good results with this approximation: https://kai-46.github.io/VisSat/ 

Cheers,

Shashank 

To view this discussion on the web visit https://groups.google.com/d/msgid/ames-stereo-pipeline-support/b57ff0ff-910c-4944-8219-664d7c8290f0n%40googlegroups.com.

[Jordan Brown]

Dear Shashank and Oleg,

Thank you both so much for these suggestions! I will try my hand at approximating the cameras and report back on how it goes.

My very best wishes and many thanks!
Jordan