Many monitoring activities of environmental processes deal with phenomena that are fuzzy. Their spatial description cannot be based on the geometry of their boundaries, because these cannot be identified. In fact, such phenomena have an uncertain thematic description and that is often the reason why their spatial extent can only be determined with limited accuracy. We will therefore present in this paper a formalism for the description of objects without fixed boundaries, i.e., objects with a fuzzy spatial extent. A method based on this formalism will be developed for the extraction of fuzzy spatial object data from digital images or from field representations in a raster format. This method can then be used for monitoring purposes when the spatial extent of fuzzy objects are identified at different epochs; the dynamics of such objects can then be obtained from these time series. Several parameters will be proposed to estimate the overlap of objects at successive epochs; with these parameters the state transitions of objects can be evaluated. Several types of state transition can be distinguished with these parameters such as shift, merge, and split of objects. These state transitions can then be combined to form the processes through which the objects evolve over time. The proposed method is applied in a coastal geomorphologic study of a barrier island in The Netherlands.
|Journal||ISPRS Journal of Photogrammetry and Remote Sensing|
|Publication status||Published - 2000|