Classification based on only spectral signatures #387
-
|
Hello! I'm writing my master's thesis on alpine nature types, and have been using the Semi-autmatic classification plugin for making a classification. I collected field data this year, and mapped ROIs that I used as training input for my supervised classification. I want to see wether these nature types have changed through the last years, so I need to use only the signatures, and not the ROIs to do classifications of the same are of previous years. I have not quite figured out how to do this, as it seems the ROIs are necessary to do a supervised classification. Is there a way around this in this plugin where I can use only the signatures from my classes, not the ROIs? Thanks alot in advance! |
Beta Was this translation helpful? Give feedback.
Replies: 1 comment
-
|
Hello @33-sl , you can export the spectral signatures as CSV file from the previous training input using the export tool Then import these signatures in a new SCP training file using the import tool (after creating the new Band set) Please note that only Spectral Angle Mapping and Maximum Likelihood algorithms work with spectral signatures (without ROIs). |
Beta Was this translation helpful? Give feedback.
Hello @33-sl ,
you can export the spectral signatures as CSV file from the previous training input using the export tool
https://semiautomaticclassificationmanual.readthedocs.io/en/latest/export_signatures_tab.html
Then import these signatures in a new SCP training file using the import tool (after creating the new Band set)
https://semiautomaticclassificationmanual.readthedocs.io/en/latest/import_signatures_tab.html#import-library-file
Please note that only Spectral Angle Mapping and Maximum Likelihood algorithms work with spectral signatures (without ROIs).