An Approach To Representation Of Region-Region Spatial Relations Based On Spatial Cognition And Its Application To Spatial Query

The spatial relations have always been the basic theory and research focus in GIS, and have been extensively applied in spatio-temporal data mining, the integration and update for multi-source/multi-type/multi-scale spatial data etc. After reviewing the existing documents and the recent progresses in related fields to spatial relations, this thesis points out some problems that are still open and one of them is that the existing researches on spatial relations are only focus on the model of spatial relations but not consider the spatial cognition. Therefore, this dissertation takes area objects as example, tries to find the model or method to describe the spatial relations based on spatial cognition. The researches in this dissertation are summarized as follows:(1) Spatial relations between spatial entities can be obtained by two means. One is computed and obtained by spatial data, and the other is obtained by human spatial cognition. So, they are named as metric spatial relations and natural-language spatial relations, respectively.(2) A method of integrated representation of spatial relations between regions is putted forward, which utilizes a strategy of decomposition and combination, spatial relations are decomposed into topological relations, direction relations and distance relations. Further, local directional and local distance relations are defined based on two adjacent intersection components (i.e. local topological relations). On this basis, the combination of these three local relations can be obtained with an ordered set of local spatial relations. Finally, the implementation of this integrated representation is employed to demonstrate the validity of the proposed approach.(3) Research on the transformation between Metric spatial relations and Natural-language spatial relations. Metric spatial relations are first depicted by combining topological predicates with metric parameters, where a mapping is built between the topological predicates and the metric parameters. Second, natural-language spatial relations are represented by combining topological predicates with quantifiers; they are vague quantifiers and critical quantifiers. Finally, the association between metric parameters and quantifiers is further illustrated and analyzed by a questionnaire.(4) Research on the representation of three objects spatial relations. We consider the projective relation as the expansion in the spatial relations, and make a further clear for the projective relation based on spatial cognition.Finally, the paper summarizes the findings, and some issues are highlighted for further investigations.