基于先验信息稀疏恢复的非均匀样本检测方法

针对训练样本被干扰目标污染导致空时自适应处理（STAP）目标检测性能下降的问题，提出了一种基于先验信息稀疏恢复的非均匀样本检测方法。该方法首先采用欠定系统局灶算法恢复待检测单元的稀疏表示系数，然后利用机载雷达系统参数等先验信息离线设计“稀疏滤波器”，并采用其滤除稀疏表示系数中的目标及“伪点”的影响，进而估计杂波协方差矩阵，最后与广义内积（GIP）方法结合，根据新的检测统计量来剔除被污染的样本。仿真分析表明，与传统GIP方法相比，该方法能够有效地检测出被干扰目标污染的训练样本，提升了STAP在非均匀环境下的目标检测性能。

Non-homogeneous Training Sample Detection Method Based on Sparse Recovery with Prior Information

For the degradation of target detection performance in space-time adaptive processing (STAP) due to non-homogeneous training samples contaminated by target-like signals,a non-homogeneous training sample detection method based on prior information and sparse recovery is proposed.The sparse representation coefficient of cell under test (CUT) is recovered using focal underdetermined system solver (FOCUSS).A "sparse filter" is constructed based on radar system parameters.The target and "pseudo point" signals are filtered out by " sparse filter",and the clutter covariance matrix is estimated.The generalized inner product (GIP) method is integrated to eliminate the contaminated training samples.Simulation analyses show that the proposed method can effectively eliminate the contaminated training samples and improve the target detection performance of STAP in non-homogeneous environment.