Recognition of Radar Target Based on Kernel Projection Subspace using Range Profiles

Abstract

A novel subspace method of radar target recognition, which is named as kernel projection subspace (KPS) method, is proposed in this paper, in which geometric relations are preserved according to prior class-label information and nonlinear variations of range profiles are represented by nonlinear mapping. So, the KPS can be used to extract target feature for enhancing local within-class relations. The experimental results of four kinds of planes demonstrate the efficiency of approach proposed in this paper.