I’d like to segmentate/classify the land cover from my interferogram’s coherence. I know that with thresholding I can set the values and interpret them with different colors for example high coherence is urban, medium is agricultural, low is forest etc. like here mentioned: http://lib.tkk.fi/Diss/2013/isbn9789526054162/isbn9789526054162.pdf
But I’m interested in does snap contain some built in classification method like Maximum likelihood, or fuzzí clustering? Or what kind of software can I use for this?
Thanks for answering!
That’s a good paper by the way @mengdahl would approve.
You could do you own thresholding with band maths. SNAP currently only has the unsupervised K-means and EM Clustering found in the Raster -> Image Analysis menu and the Wishard classifier in the Radar -> Polarimetric menu. Soon by May, we should have other supervised methods including Maximum Likelihood, SVM and Random Forest.
Thank you so much for answering, I could apply K-means custering for getting a good result.
Do you have a handbook for SNAP, which describes the main algorithms? I know in Help there is some useful information about what happens in different steps (TOPSCoregistration, Interferogram phormation etc.).
I would like to know the TOPS Coregistration works with crosscorrelation computing or baseline evaluation…