Nate Andrysco

Urban Grammar
A continuation of the AMUSE project which deals with visualizing large urban
environments. In Urban Grammar, a 2D map of a city is parsed to detect roads, land
boundaries, and building outlines. The parsing creates a unique grammar which
describes the city. Once the information is obtained, we can then derive new cities in the
style of the original using the grammar. This process includes how to obtain the
boundary data, how parsing is conducted, and how derivations are done. It will also be
shown how cities may be edited so that a city planner may use this tool for his work.
Issues involved in editing are grammar simplification, region similarity estimation, and ngon
to m-gon mapping. Grammar simplification greatly helps reduce the number of
computations needed during derivation. Similarity estimation is an essential tool needed
in derivations and most phases of editing. N-gon mapping is used to project texture
coordinates from a region in the original city to a new, possibly deformed, region in the
edited city.