Detecting cultural features in urban SAR images for wireless applications

This paper explores the use of L- and C-band SAR data for detecting cultural features in urban SAR images. The resulting map will be used in propagation models to assist in siting antennas for wireless communication. The research starts with the classification of a dataset of Dallas, Texas into several classes based on backscattering coefficient statistics, IFSAR-derived height variation and density of the buildings and vegetation. Further identifying major road networks, airports and malls is also extremely important due to the high usage of cellular phones in those areas. Because each has a very distinctive look in a SAR image, involving linear features with low backscattering intensities for roads, large open areas for runways and malls, and also a dense concentration of buildings (stores) for malls. These targets are amenable to automatic detection. The authors describe their efforts to classify each of these cultural features. This is then compared with the known locations in the area.