摘要
In 3D urban visualization, large data volumes related to buildings are a major factor that limits the delivery and browsing speed in a web-based computer system. This paper proposes a new approach based on the level of detail （LOD） technique advanced in 3D visualization in computer graphics. The key idea of LOD technique is to generalize details of object surfaces without losing details for delivery and displaying objects. This technique has been successfully used in visualizing one or a few multiple objects in films and other industries. However, applying the technique to 3D urban visualization requires an effective generalization method for urban buildings. Conventional two-dimensional （2D） generalization method at different scales provides a good generalization reference for 3D urban visualization, Yet, it is difficult to determine when and where to retrieve data for displaying buildings. To solve this problem, this paper defines an imaging scale point and image scale region for judging when and where to get the fight data for visualization. The results show that the average response time of view transformations is much decreased.
In 3D urban visualization, large data volumes related to buildings are a major factor that limits the delivery and browsing speed in a web-based computer system. This paper proposes a new approach based on the level of de- tail (LOD) technique advanced in 3D visualization in computer graphics. The key idea of LOD technique is to gener- alize details of object surfaces without losing details for delivery and displaying objects. This technique has been successfully used in visualizing one or a few multiple objects in films and other industries. However, applying the technique to 3D urban visualization requires an effective generalization method for urban buildings. Conventional two-dimensional (2D) generalization method at different scales provides a good generalization reference for 3D ur- ban visualization. Yet, it is difficult to determine when and where to retrieve data for displaying buildings. To solve this problem, this paper defines an imaging scale point and image scale region for judging when and where to get the right data for visualization. The results show that the average response time of view transformations is much de- creased.
基金
Project 40471109 supported by the National Natural Science Foundation of China
