Many blank areas in the generated DEM when processing KH-4B data

[Zhangyu Sun]

Dear,

When I tried to use ames stereo pipeline to process the Corona KH-4B data, I found the generated DEM has many blank areas with no value. Is that normal for processing declassified data? What could be the reason causing such results? Is there any way to improve such results to obtain a more complete DEM?

[Oleg Alexandrov]

Things seem to be degrading on the left end, and it is likely because of a combination of poorly modeled camera (which our tools don't do a good job with) and large search range. 

If you have a prior DEM of that area, such as Copernicus (https://stereopipeline.readthedocs.io/en/latest/next_steps.html#choice-of-initial-guess-terrain-model), you can mapproject both the left and right image that you have on this DEM and compare them. Likely they are more different on the left. 

Then, you can run our tools with mapprojected images (https://stereopipeline.readthedocs.io/en/latest/next_steps.html#running-stereo-with-mapprojected-images). It is very important to follow all steps there, including using dem_geoid to adjust the height of an input DEM, and apply some blur to that input DEM, as your area is very mountainous. Then mapproject at the same resolution and run stereo and point2dem. 

For stereo, one has to be mindful of the search range being large if the images differ a lot. See https://stereopipeline.readthedocs.io/en/latest/next_steps.html#dealing-with-long-run-times-and-failures.

For point2dem, it is strongly suggested to find the error image and examine it. It will likely be high at the left end, which again comes from our models needing more improvement there. We have tools to refine intrinsics (https://stereopipeline.readthedocs.io/en/latest/bundle_adjustment.html#using-the-heights-from-a-reference-dem) but these take work and experimentation.

--
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/1dd23663-c918-43ad-aad9-ccaea2fd8578n%40googlegroups.com.