Aggregation of 3D Buildings Using a Hybrid Data Approach

Abstract

With the increasing availability of large-area 3D city models, there is also a rising need for cartographic generalization algorithms to prepare this data so as to improve both the performance and the perceivability of their visualizations, especially in a mobile-mapping context. Most research has focused on the geometric simplification of single 3D buildings, which can only reach a certain degree of abstraction until building-specific geometric constraints are violated. To obtain further levels, an aggregation of building groups is necessary. The previously published building generalization approach based on cell decomposition is therefore extended to use morphological operations on a raster representation of the initially vectorial data. By repeatedly applying dilation and erosion, the space between buildings closes, which boosts the importance of the already available building blocks in these areas so that they are now included in the generalized model. Even though the results are encouraging from a technical point of view, it leads to a fundamental discussion of how the roofs of aggregated 3D buildings should appear.