In this paper we present initial results from an automated terrain classification algorithm that utilizes the information collected by an interferometric synthetic aperture radar system. It is shown that by combining radar cross section imagery with height maps additional information concerning terrain types can be extracted automatically, with the height information differentiating forests from other terrain classes and the radar cross section information differentiating field types. When tested on the same data it was trained on classification accuracy of 94 percent to 100 percent are shown. Similar results are generated when tested on a different data set, although to date this has only been determined with visual comparisons. The largest problem with the algorithm is the use of absolute height information that confuses high fields with forests. Work is on-going to develop relative height measures to improve the robustness of the algorithm.
展开▼
