Mapping Built-up Areas from Multitemporal Interferometric SAR Images - A Segment-based Approach

Abstract

Automatic mapping of built-up areas from a multitemporal interferometric ERS-1/2 Tandem dataset was studied. The image data were segmented into homogeneous regions, and the regions were classified as built-up areas, forests, and open areas using their mean intensity and coherence values and additional contextual information. Compared with a set of reference points, an overall classification accuracy of 97 percent was achieved. The classification process was highly automatic and resulted in homogeneous regions resembling a map drawn by a human interpreter. The feasibility of the imagery for dividing built-up areas further into subclasses was also investigated. The results suggest that low-rise areas, highrise areas, and industrial areas are difficult to distinguish from each other. On the other hand, a correlation between the building density, the proportion of land covered with buildings, and intensity/coherence in the image data was found. The dataset thus appeared to be promising for classifying built-up areas into subclasses according to building density.