Radar Emitter Recognition Based on LE-SVDD Classifier

For existing radar emitter recognition method, the performance declines sharply with decreasing of training sample numbers. To overcome this problem, a LE-SVDD classifier is proposed and applied in radar emitter recognition. In the LE-SVDD classifier, Laplacian Eigenmaps (LE) algorithm, which belongs to manifold learning field, is employed to improve SVDD classifier. And meanwhile, the strong generalization capability of SVDD classifier is inherited in it. So it is suitable for radar emitter recognition under a small-scale training sample set. The experimental results show that: when the number of training sample is small, higher radar emitter recognition accuracy can be achieved by the proposed method.