Automatic Target Recognition For the TOPCAT System.

摘要

We have developed a proposed sensor system for the TOPCAT anti-armor system. It analyzes cross-sectional measurements of the underlying terrain, detecting targets by their distinctive shape. Sparsely-sampled point from a single scan line are fed as input to a Parzen estimator, which makes the target/no-target decision. Morphological operators then remove spurious noise points, producing large blobs over the detected target region. Results are shown, using the GTVISIT simulation system to produce test data. ATR, Automatic target recognition, TOPCAT, Statistical pattern recognition, Parzen classifier.