Efficient Retrieval of 3D Building Models Using Embeddings of Attributed Subgraphs

University of Bonn, Technical Report number CG-2011-2, Aug. 2011
 

Abstract

We present a novel method for retrieval and classification of 3D building models that is tailored to the specific requirements of architects. In contrast to common approaches our algorithm relies on the interior spatial arrangement of rooms instead of exterior geometric shape. We first represent the internal topological building structure by a Room Connectivity Graph (RCG). Each room is characterized by a node. Connections between rooms like e.g. doors are represented by edges. Nodes and edges are additionally assigned attributes reflecting room and edge properties like e.g area or window size. To enable fast and efficient retrieval and classification with RCGs, we transform the structured graph representation into a vector-based one. We first decompose the RCG into a set of subgraphs. For each subgraph, we compute the similarity to a set of codebook graphs. Aggregating all similarity values finally provides us with a single vector for each RCG which enables fast retrieval and classification. For evaluation, we introduce a classification scheme that was carefully developed following common guidelines in architecture. We finally provide comprehensive experiments showing that the introduced subgraph embeddings yield superior performance compared to state-of-the-art graph retrieval approaches.

Images

Download Paper

Download Paper

Bibtex

@TECHREPORT{wessel-2011-Efficient,
       author = {Wessel, Raoul and Ochmann, Sebastian and Vock, Richard and Bl{\"u}mel, Ina and Klein, Reinhard},
        title = {Efficient Retrieval of 3D Building Models Using Embeddings of Attributed Subgraphs},
       number = {CG-2011-2},
         year = {2011},
        month = aug,
  institution = {University of Bonn},
     abstract = {We present a novel method for retrieval and classification of 3D building models that is tailored to
                 the specific requirements of architects. In contrast to common approaches our algorithm relies on
                 the interior spatial arrangement of rooms instead of exterior geometric shape. We first represent
                 the internal topological building structure by a Room Connectivity Graph (RCG). Each room is
                 characterized by a node. Connections between rooms like e.g. doors are represented by edges. Nodes
                 and edges are additionally assigned attributes reflecting room and edge properties like e.g area or
                 window size. To enable fast and efficient retrieval and classification with RCGs, we transform the
                 structured graph representation into a vector-based one. We first decompose the RCG into a set of
                 subgraphs. For each subgraph, we compute the similarity to a set of codebook graphs. Aggregating all
                 similarity values finally provides us with a single vector for each RCG which enables fast retrieval
                 and classification. For evaluation, we introduce a classification scheme that was carefully
                 developed following common guidelines in architecture. We finally provide comprehensive experiments
                 showing that the introduced subgraph embeddings yield superior performance compared to
                 state-of-the-art graph retrieval approaches.}
}