Distance based range profile classification techniques for aircraft recognition by radar - a comparison on real radar data
conference paper
Aircraft identification is essential in any air-defence scenario. Without a robust classification capability no effective threat evaluation can be performed. A prominent aircraft recognition technique is based on the exploitation of a one-dimensional image of a target, a range profile. In this paper, we employ four different classification techniques, all based on shift invariant distances, and a method to compare them. Two of the techniques are based on Radial Basis Functions for which a novel technique to optimize the number of free parameters is presented. The application is on real radar data, where a true separation between train- and test profiles is accomplished. The classifìcation results are encouraging. As an example, a qualitative statement is given about the best of the four classifiers to be used in combinations of two scenarios and four applications.
TNO Identifier
94840
Publisher
Advisory Group for Aerospace Research and Development (AGARD ; NATO)
Source title
AGARD Conference Proceedings on Tactical Aerospace C3I in Coming Years - Papers presented at the Mission Systems Panel 3rd Symposium held in Lisbon, Portugal from 15th May to 18th May 1995
Place of publication
Neuilly-sur-Seine
Pages
17/1-17/7