Aspects of automatic target recognition with a two-frequency millimeter-wave SAR

In the presentation it will be shown that the classification approach based on polarimetric features supplemented by geometrical information shows promising results using turntable measurements of one of the targets as reference data and airborne measurements of a set of targets as test data. Furtheron the results using the split measurement data sets (split into reference and test sets) will be discussed in detail. The two approaches are compared to each other and their respective advantages and disadvantages summarized.