I. Introduction
Urban environments represent one of the most challenging areas for remote sensing analysis due to high spatial and spectral diversity of surface materials. Typical urban surface types include a wide range of roofs, roads, sidewalks and parking lots of variable age, quality and composition. Further complicating the urban landscape are bare soil, vegetation, and other landscaping elements, creating a spectral diversity that far exceeds natural environments. This complexity, along with three-dimensional surface heterogeneity, creates a particularly challenging mapping environment for urban areas [1]–[4].