ABCRaster stands for Accuracy assessment of Binary Classified Raster. It is a package for performing validation, accuracy assessment, or comparing binary classified rasters (.tiff) versus a reference (.shp). Primary use case is to compare flood maps encoded as (1,0) in tiff file format against a reference vector from CEMS.
MIT License
0
stars
0
forks
source link
Harmonize Confusion Map Extents between Raster and Vector Inputs. #6
Confusion map extents generated from raster and vector references are different and should be harmonized. For shapefile references, output confusion maps are generated with the same extent as the input classified raster (with the outside of reference bounds as no data). Whereas, the from Geotifs, output confusion map extent is the intersection of the input and reference raster. The latter being the preferred behavior.
There is the option clip2bbox, which allows clipping the vectorized reference data to its own bounding box. However, this needs to be improved and the convex hull instead of the bounding box should be used.
Confusion map extents generated from raster and vector references are different and should be harmonized. For shapefile references, output confusion maps are generated with the same extent as the input classified raster (with the outside of reference bounds as no data). Whereas, the from Geotifs, output confusion map extent is the intersection of the input and reference raster. The latter being the preferred behavior.