I was wondering if someone has delt with this kind of issue before:
I want to delineate fragments based on neighboring pixels of the same class (forest) with a distance threshold of 3 pixels (which is biologically significant for my specie of interest).
My concern is, see the attached image for an example, that sometimes these fragments are actually corridors, and often corridors and actual fragments are grouped into a same fragment because of their proximity.
I'm wondering if there is a way to distinguish the corridors from the fragments based on shape, number of surrounding pixels, etc..?
For instance, in the following box, possible corridors are indicated by the red boxes, and fragments by the green ones.
I have access to QGIS and R.