Recognition of Polymorphic Patterns in Parameterized Graphs for 3d Building Reconstruction

An approach for the recognition of polymorphic patterns by subgraph isomorphism computation of parameterized graphs will be presented. Parameterized graphs (short: p-graphs) are extensions of undirected graphs by parameter vectors at the nodes and edges. We will deene p-graphs and basic concepts of subgraph isomor-phism computation for p-graphs. A bottom-up algorithm for p-subgraph isomorphism computation according to a given search graph and a template graph will be described. Since the enumeration of all induced p-subgraphs require tremendous eeort, we will propose four pruning mechanisms in order to reduce the size of the search space. The computed subisomorphisms provide useful background knowledge for 3D building reconstruction in order to alleviate the occurring ambiguities, namely occlusions, inverse mapping, and noise. Using an extensive suburban scene it will be shown how polymor-phic patterns can be recognized. Finally, we will estimate and discuss an upper bound complexity of the algorithm according to the application.