Image-mining for solving spatial and spatio-temporal uncertainty

摘要

Commonly, spatial uncertainty is seen as a nuisance, although it is inherently present in real world objects. This is largely due to more complicated analysis and data collection procedures required to overcome the uncertainty. This paper shows that with relatively simple means some of the problems can be addressed using modern statistical methods. On the basis of a well-defined mathematical backbone, a proper reasoning mechanism is formulated. Next, image-mining methods are described for space time data collection. Errors in the positional accuracy of polygons are propagated into an environmental cost model. Finally, an optimal spatial sampling strategy is defined for areas segmented into fuzzy regions. We conclude from this paper that spatial uncertainty can be seen as an asset in better modeling and analyzing spatial and spatio/temporal phenomena.