I'm running QGIS 2.16 on Windows 10.
I'm looking to perform an unsupervised K means classification on a geoTiff image I have from a multispectral camera surveying sugar cane fields.
Alas, the documentation for the algorithm in QGIS needs work done to make the code user friendly to use in the Python console.
Parameters Input Image [raster]
Available RAM (Mb) [number] Default: 128
Validity Mask [raster] Optional. ## I don't know what this means.
Training set size [number] Default: 100 ## Or this
Number of classes [number]
Maximum number of iterations [number]Default: 1000 ##Don't know either
Convergence threshold [number] Default: 0.0001
Output Image [raster] ## I imagine it's the layer with the classification done Centroid filename [file] (Not exactly sure, I know what centroids are but ##those work for polygons in vector files not raster files).
I'm thinking maybe it's because I don't have the OrfeoToolbox installed but I´m positive I do because I downloaded and installed QGIS 2.16 along with GRASS and the OsGeo4w packages.