摘要
This paper provides a brief introduction to the methods for generating fuzzy categorical maps from remotely sensed images (in graphical and digital forms).This is followed by a description of the slicing process for deriving fuzzy boundaries from fuzzy categorical maps,which can be based on the maximum fuzzy membership values,confusion index,or measure of entropy.Results from an empirical test preformed in an Edinburgh suburb show that fuzzy boundaries of land cover can be derived from aerial photographs and satellite images by using the three criteria with small differences,and that slicing based on the maximum fuzzy membership values is the easiest and most straightforward solution.This,in turn,implies the suitability of maintaining both a crisp classification and its underlying certainty map for deriving fuzzy boundaries at different thresholds,which is a flexible and compact management of categorical map data and their uncertainty.
This paper provides a brief introduction to the methods for generating fuzzy categorical maps from remotely sensed images (in graphical and digital forms). This is followed by a description of the slicing process for deriving fuzzy boundaries from fuzzy categorical maps, which can be based on the maximum fuzzy membership values, confusion index, or measure of entropy. Results from an empirical test preformed in an Edinburgh suburb show that fuzzy boundaries of land cover can be derived from aerial photographs and satellite images by using the three criteria with small differences, and that slicing based on the maximum fuzzy membership values is the easiest and most straightforward solution. This, in turn, implies the suitability of maintaining both a crisp classification and its underlying certainty map for deriving fuzzy boundaries at different thresholds, which is a flexible and compact management of categorical map data and their uncertainty.
