Detection and extraction of buildings from interferometric SAR data

In this paper, we present a complete procedure for the extraction and chara cterization of building structures starting from the three-dimensional (3-D ) terrain elevation data provided by interferometric SAR measurements. Each building is detected and isolated from the surroundings by means of a suit ably modified machine vision approach, originally developed for range image segmentation. The procedure is based an a local approximation of the 3-D d ata by means of best-fitting planes. In this way, a building footprint, hei ght and position, as well as its description with a simple 3-D model, are r ecovered by a self-consistent partitioning of the topographic surface recon structed from interferometric radar data.