Qualitative Representation And Retrieval Application On 3D Topological Relationships

Three-dimensional spatial relationships in Geographic Information System(GIS) has a broad application prospect, it is an important topic at the forefront of GIS based theory. The range of GIS includes not only the limits of the atmosphere and the outer space of the earth, also includes the surface of the earth’s space and interior space of the earth. Its space objects include atmospheric environment, building roads, pit slope, mine engineering, rock formation, etc. These objects have complex spatial relationships, which exist in the form of complex three-dimensional spatial entities. The simulation description and analysis of 3D GIS system is dependent on the three dimensional spatial relations, so 3D topological analysis and 3D space analysis have already become the focus of attention of the researchers.The existing spatial reasoning model is based on two-dimensional object in most cases, and for two-dimensional spatial relationships has been very mature. By comparison, for the study of the three-dimensional spatial relationship is less. The existing research is mainly focused on 3D topological relationships. Three-dimensional topological relation research focuses on how to effectively define all kinds of spatial entities. Compared with the two-dimensional case, the spatial entities of the 3D space is more complicated. The definition and description of the three-dimensional entity is very difficult. The existing 3D space relationship model generally studies the 3D entity of the relatively simple structure. For some complicated structure of the three-dimensional entities, because of the data of representing the entity has complex structure, their spatial relationship is often difficult to describe and deal with.It is the well-known model for the 4I model and 9I model in the conventional three-dimensional model of the spatial relationship. 4I model is a binary topological relations model, it makes the each space entity as the composition of the set of boundary and internal. 9I model is based on 4I model, and it increases the external of spatial entities. The model is defined by the intersection of the contents of two spatial entities of boundary, external and internal to determine the topological relationship between the two entities. Both models have the relatively strong ability to distinguish the topological relationship.At present although some 3D spatial relation model have been proposed, but the similarity of the study of the 3D spatial relations did not carry out. The retrieval of 3D spatial relations cannot be achieved as the lack of similarity measurement of 3D spatial relationships. To this end, this paper proposes a new 3D topological relation model--the 3D entity connected graph model. In this model, if two 3D entities exist public parts, it will establish a connection relationship between two 3D entity and form the graph model. We put forward a 3D spatial relations similarity measure method as the basis on this model.The retrieval work of this paper includes 3D entity retrieval and the 3D spatial relations retrieval. The 3D entity retrieval is by extracting characteristic to represent 3D model, then by comparing the characteristic values, completing similarity measure 3D entities. The 3D entity retrieval improves and applies to the mature technology of the 3D model retrieval. The 3D spatial relation retrieval is based on the 3D entity retrieval, and uses the 3D spatial topological relationship model that we propose. In this article we finally have achieved to the retrieval of 3D scene as the consisting of two 3D entities, and we carried out a simulation experiment.The work of this paper mainly has the following several aspects:1. It is simply introduced the relevant theoretical knowledge of 3D model retrieval and it provides theoretical basis for 3D spatial relations retrieval.2. The paper has an overview for the current research status existing 3D topological relationships model. And we analyze the application prospect and the corresponding lack. This paper highlights kinds of 3D topological model. The paper analysises and discusses the research status and the problems of these types of models. Then according to the existing theory and the three-dimensional entity connected relationship, the paper proposes the 3D entity connected graph model. We give the algorithm of the building three-dimensional entity connected graph. According to this algorithm we get a 3D topological relationships graph of two entities.3. The paper briefly introduces the graph spectral theory, then we give a similarity measurement algorithm of 3D topological relationships graph of two entities that bases on the graph spectral theory and 3D topological relationships graph. Then we construct the similarity measurement formula of 3D entity model that bases on the theory of 3D model retrieval. Finally, we combine the similarity measurement of 3D topological relationships graph and give the similarity measurement algorithm of two 3D entity spatial relationships.4. We built the test library and tested, it was effective that it validated the presented method by the contrast experiment.