Research On Model And Query For Directional Relation

Spatial database has gained more and more attentions in recent years, it is a interdisciplinary subject. Spatial relation representation and procession is important in numerous application domains such as cartography, computer-aided design, image and multimedia databases, spatial and geographic applilcations. Spatial relations mainly include topological relation, directional relation and distance relation. Directional relation describes the order between objects. it is a important research area on spatial relations. Before most papers concentrate on topological relation and distance relation. Directional relation has gained more and more attentions in recent years.Spatial relation is a relation with some property between spatial objects. Drectional relation is one kind of order relations. It is also a basic relation in spatial relations and a hot area on research. Directional relations describe the order between spatial objects in space using words on directions.This paper mainly researched two aspacts of directional relation in database: directional relation model and query, generalized and compared theories and methods which are achieved. The main contributions and results included in the thesis are as follows:Firstly, the thesis summarizes and analyses directional relation models which existed so far, compares the advantages and disvantages among all kinds of directional models, And then presents a new directional relation model-star-based model according to their defects, features of directional relation and requirements of applications. This model has some properties, such as formalization, cognition, multi-scale, compatibility and so on.Secondly, directional queries that have been given are studied. A new method based on R-tree for diretional query ia proposed by combining with the new directional relation model and requirements of applications. At the same time, a calculation for similarity based on directional matrix is given. The method improves directional relation matrix for formalized representation, and implements calculation for similarity on the condition that the distance is constant for the directional relation matrix.Thirdly, at last the experiment platform is built up based on the new directional relation model and the new method for directional query, and then the performance and the feature are probed. The theory analysis and evaluation based on experiment prove that directional query based on R-tree improves efficiency.