系统接收数据的杂噪协方差矩阵完整地包含了杂波和噪声信息 ,将其做特征分解 (EVD)而得到的特征空间 ,可将杂噪能量按序排列 ,且各特征矢量相互正交且统计独立 .基于特征空间作自适应处理 ,特别是降维自适应处理 ,便于探寻理论上的最佳方案 .本文分别讨论了互谱法 (CSM)、最小功率特征对消 (MPE)以及最小范数特征对消(MNE) ,从特征结构阐明了三种方法的内在关系 ,比较了它们的杂波抑制性能 . The noise-based covariance matrix of the received data of the system completely contains clutter and noise information. The eigen-space obtained by EVD can arrange the noise energy in order, and the eigenvectors are orthogonal to each other Statistical independence.It is easy to search for the best theoretical scheme based on the feature space for adaptive processing, especially the dimension reduction adaptive processing.In this paper, we discuss the correlation between the cross spectrum method (CSM), minimum power feature elimination (MPE) (MNE), the inherent relationship between the three methods is elucidated from the feature structure, and their clutter suppression performance is compared.