Research On The Observer-Orientation-Based Direction Relation Model And Query Method

Directions are frequently used as select conditions in spatial queries. The observer-orientation-based direction relations are very important to many fields, for example multi-media database, artificial intelligence, virtual reality and city tour. But the observer-orientation-based model and query methods of spatial direction relations are less researched at present, and it is still not have an unified theory.Firstly, the basic theory of spatial relations and available models are discussed. There is a summary about quantitative models and qualitative models. Basing on the traditional projection-based direction relation model, an observer-orientation-based precise model are defined, which has a definition to observer-orientation-based direction relations and a new division to the direction areas according to observer's visual characters. Two kinds of direction predicates are listed and their completeness are proofed.Secondly, aming at the shortcuts of the traditional precise model cannot express the uncertainty in direction relations very well, the fuzzy set theory in math is introduced into the observer-orientation-based model. A new observer-orientation-based fuzzy model are established and an array of membership functions are proposed for every fuzzy direction area. Their validity and feasibility are also be verified.Finally, aming at the shortcuts of the traditional range query method, which means the low efficiency and the need of filter, when the rage query method is applicated in observer-orientation direction relations, basing on the observer-orientation-based model, an improved R*-tree index and two query methods, observer query method and fuzzy model query method, which based on the observer orientation are proposed. Their I/O costs are analysed and experiments on diffirent parameters are completed to verify their performences.