Sound and dynamics of targets - Fusion technologies in radar target classification

The challenge of modern sensor systems is besides the tracking of targets more and more their classification. The knowledge of the target class has significant influence on the identification, threat evaluation and weapon assignment process of large systems. Especially, considering new types of threats in Anti Asymmetric Warfare the knowledge of a target class has an important drawback. Also the target class is used to optimize track and resource management of today's agile sensor systems. A technology is presented that fuses different classifiers to decide between persons, tracked vehicles, wheeled vehicles, helicopters, propeller aircrafts and clutter for a 2 dimensional, electronically scanned radar system. A first classifier analyses the Doppler sound of the target to decide its target class. Therefore, a cepstrum based feature extractor and a Hidden Markov Model (HMM) is applied. Similar to techniques that have been well proven in speech and image recognition, the time-varying nature of radar Doppler data is exploited. A second type of classifier extracts dynamic features of the target found by the underlying target tracking methodology. Through a fusion of these classifiers a highly reliable classification result is established.