An Approach to 3D Building Model Retrieval Based on Topology Structure and View Feature

This paper proposes an approach to 3-D building model retrieval based on a topology structure and view feature using two filter steps to finish the building model retrieval. First, the distance transformation method is used to extract the topological structure of 3-D building models. The obtained skeleton points of 3D building models are classified by our concentric sphere method. The 3-D building models are then filtered for a first time based on the distribution features. Second, 3-D building models are projected onto 2-D images from the viewpoints of skeleton points. The scale-invariant feature transform algorithm is used to extract feature points from 2-D projection images, and the idea of the bag of feature method is used to compare the projection images. The 3-D building models are then filtered for a second time. Finally, we can get the retrieval result (the matched models). This method first carries out the matching of the overall structure and then carries out the matching of local details. This conforms to the cognitive style of human. In addition, the experiment results show the effectiveness of our method.