For a project, we are processing interferometric TerraSAR-X data. The pipeline generally works like a charm, but we would like to use a more sophisticated coherence/phase estimator than boxcar (or Goldstein). Is there any way to carry out this step externally, but to reintroduce the resulting coherence/phase images to SNAP for further processing?
To be more precise, the workflow would conceptually look something like this:
- Image coregistration → SNAP
- Flat earth (terrain) removal → SNAP
=> EXPORT COREGISTERED AND CORRECTED SLC IMAGES
- Phase/coherence estimation (e.g. using NL-InSAR or some deep learning-based solution) → external
=> IMPORT PHASE/COHERENCE IMAGES BACK TO SNAP
- Phase-to-height conversion → SNAP
- Geocoding → SNAP
and so on.
Or would there be a completely different solution to my desire? The whole point is to get better phase/coherence estimates into the system…