A 3-D city model data structure and its implementation in a relational database

With the development of modern cities, 3-D spatial information systems (SIS) are increasingly required for spatial planning, communication systems and other applications. The geometric information to be used in a spatial information system usually includes two types: vector data (such as buildings, traffic ways, waterways, trees, DTM, etc.) and raster data (such as orthophotos, original images from aerial or still video cameras, etc.). Considering the availability and advantages of relational database technology, it is an important task to develop a data structure which integrates vector and raster data. In this paper, a self-developed data structure (V3D) is presented. The basic geometric element is the point. Points are used to express faces and line segments. Hierarchically, a 3-D object consists of faces, a face consists of ordered polylines, a polyline consists of ordered line segments, and a line segment consists of points. Thus the geometrical and the topological information of an object are defined. Once some attributes are attached, a geometrical object becomes an entity class. A total of four classes are defined: 3-D body, surface, line and point. The DTM is considered a separate entity class because some applications require particular processing of DTM. In addition, the images taken from aerial or still video cameras are attached to the faces as special attributes. We will present the data structure together with the implementation in a relational database. Also, we will report about a prototype system, CC-SIS (CyberCity Spatial Information System).