噪声中目标距离剖面像识别的修正特征子空间法

以噪声背景下高分辨雷达目标的距离剖面像为测量矢量 ,提出了基于修正特征子空间而进行特征抽取的目标识别方法。对每类目标 ,由训练样本矩阵经奇异值分解 (SVD) ,建立各自的特征子空间 ,并提取相应的降噪算子。对未知目标 ,其距离剖面像在各特征子空间中的投影 ,经相应降噪算子作用后 ,以能量最大为准则进行识别。模拟实验结果表明了所提方法的有效性

A Modified Eigen-Subspace Method for Target Recognition Using Noisy Range Profiles

A novel approach of noisy range profile recognition is proposed in this paper. This approach executes singular value decomposition (SVD) on matrix formed by target's noisy range profiles of high-resolution radar (HRR) in different aspect angles. After SVD, an eigen-subspace and a noise-reduction operator are obtained for each class of target. The range profile of an unknown target is mapped into each subspace and transformed by the corresponding operator. The paper uses energy maximum of the transforming as rule for target classification. Results of simulated experiments show the efficiency of the approach in noisy target recognition.