Automated rapid mapping with interferometric synthetic aperture radar data

Abstract

This paper describes automated algorithms that generate map products in a rapid fashion with interferometric synthetic aperture radar (IFSAR) data set. The input IFSAR data set consists of backscatter, elevation and correlation images and the automatically extracted map elements include land-use polygons, transportation networks, buildings/man-made structures, bare soil elevation contours and hydrology networks. A calculator computes five feature images from the IFSAR data set and the five feature images serve as inputs to a hybrid unsupervised classification algorithm which segments the scene into five classes. The five classes include fields, trees, urban area/man-made structure, water and unknown. The classification results allows a bare soil terrain model to be extracted from the IFSAR terrain model using a novel algorithm. The difference between the bare soil elevation and the original IFSAR elevation can be used in conjunction with the classification map to identify and estimate the heights of buildings, man-made structures and trees. Elevation contours and hydrology networks derived front the bare soil model can be shown to be more accurate than that derived from the IFSAR model. Transportation networks are identified using a road extraction technique based on rotational energy summation. Sample map products are shown to demonstrate the efficiency and accuracy of the automated algorithms.