Increasing the classification performance of a classifier via sub-space based classifications

One of the challenging problems in modern radar applications is the task of "non-cooperative target identification" (NCTI). Under many circumstances it is highly important to find out about the class or even identity of a target under observation without a cooperative support of this object. Especially for air-targets, but also for targets on the ground, presently at many places in the world classification methods are investigated, which are based on radar imagery of these targets. In case of a one-dimensional image we speak of "high-range resolution"-profiles (HRR-profiles) of the object under consideration, but also two dimensional images gained via SAR or ISAR methods could be used. The following paper deals with the identification of air-targets using HRR-profiles.