Generalization of 3D city models with MapReduce

Abstract
In this paper we introduce a 3D city model generalization system based on the Hadoop platform using MapReduce framework. The input datasets in CityGML is first divided into separate buildings which are handled by Mapper for simplification. The Mapper simplifies the building by welding the vertices close enough and removing redundant faces. At last the simplified buildings are aggregated into a complete CityGML file ready to use. The experiment shows the effectiveness of the generalization algorithm and the practicability of processing large scale geographic data with MapReduce framework and the Hadoop platform.