Resolving overlaps/slivers by weight
I'm currently labelling a satellite image of the coast. There are four labels: vegetation, urban, sand, water. To help delineate the correct polygon boundaries, i.e., between water and sand, we have 5 users labelling the image using polygons.
Once each user has mapped the image, the count overlaps tool is used to obtain the following attribute table for each label:
Count refers to many users out of 5 labelled this area as X. For example in the above attribute table for sand.shp, polygon 4027 represents an area labelled as "sand" by 1/5 users whereas polygon 4029 represents an area labelled as "sand" by 2/5 users.
Using the "count overlap" polygons, I wish to resolve any remaining slivers/overlaps using the following criteria.
Overlaps between labels with count values > 2: select the class with the highest overlap count value (e.g., if 3/5 users labelled it as water and 2/5 sand, then label the overlap as water). If the overlap value is equal (e.g., 2/5 for sand and 2/5 for water) then select according to this weight "urban > vegetation > sand > water".
Overlaps between labels with count values = 1: label according to this weight "urban > vegetation > sand > water".
What is the best way to go about implementing this criteria for filling in gaps?
