Centrality measures in spatial networks of urban streets

We study centrality in urban street patterns of different world cities represented as networks in geographical space. The results indicate that a spatial analysis based on a set of four centrality indices allows an extended visualization and characterization of the city structure. A hierarchical clustering analysis based on the distributions of centrality has a certain capacity to distinguish different classes of cities. In particular, self-organized cities exhibit scale-free properties similar to those found in nonspatial networks, while planned cities do not.