| Abstract |
A digital city is the basis of the Smart city. With the deepening exploration of smart applications, traditional geometric models become hard to satisfy the needs of precise space description in urban planning and management. How to add meanings to the spatial data and construct semantic models of cities has been one of questions of Geo-informatics nowadays. In this paper, we propose an automatic approach to achieve semantic annotation of 3D architecture models, which are the main components of cityscapes in the virtual environment. On one hand, all the concave and convex features on the surfaces are extracted as the clues to decompose the architecture models into different structural parts in geometry; On the other hand, the positions, shapes, sizes, and configurations of the extracted structural parts are analyzed to decide the semantic category that each of them belongs to. To verify the effectiveness of the approach, experiments have been carried out on a number of architecture models, and the semantic cognition capability of the approach is demonstrated. |
