Radar target feature extraction by wavelet transform and KCN

Feature extraction of radar target recognition using high range resolution profiles is discussed. A feature extraction scheme by using wavelet transform and Kohonen clustering network (KCN) is proposed from the points of reducing feature dimension and feature numbers to represent a target. To compare the performance of different feature extraction methods, Fisher distance measure is introduced which reveals the within-class compactness and the between-class separation. Experimental results show the effectiveness of this method.