非均匀环境中分布目标的参量广义似然比检测

 在高距离分辨率(HRR)雷达中，目标很可能跨越多个距离门。该文研究了这种分布目标的参量自适应检测。其中，主、辅数据中的干扰信号用随机空域协方差矩阵的向量自回归模型表示。随后，分别根据贝叶斯1步参量广义似然比(B1S-PGLRT)和贝叶斯两步参量广义似然比(B2S-PGLRT)检测准则推导了对应的检测器。前者没有闭式解而后者和经典的参量自适应匹配滤波器(PAMF)具有相似的检测结构，并使用了空域协方差矩阵的最大后验(MAP)估计代替了最大似然估计(MLE)。同时，还给出了B2S-PGLRT的归一化形式。最后，分析了贝叶斯参量检测器的运算步骤和运算复杂度，并通过蒙特卡洛仿真评价了它们的检测性能。结果表明：当训练数据不足时，贝叶斯框架下的参量匹配滤波器比广义似然比性能更好。

Parametric Generalized Likelihood Ratio Tests for Distributed Target in Heterogeneous Environment

In High Range-Resolution (HRR) radar, a target extends probably more than one range bins and the parametric adaptive detection is studied in this paper for such distributed target, where the disturbances in both primary and secondary data are represented by a vector autoregressive model with random spatial covariance matrix. Subsequently, the corresponding detectors are derived according to Bayesian one-Step Parametric Generalized Likelihood Ratio Test (B1S-PGLRT) and Bayesian two-Step Parametric GLRT (B2S-PGLRT) decision rules. However, the former leads to no close formulation and the latter has the similar detection architecture with the classic Parametric Adaptive Matched Filter (PAMF), using Maximum A-Posteriori (MAP) estimator instead of Maximum Likelihood Estimator (MLE), of the spatial covariance matrix. Meanwhile, the normalized version for B2S-PGLRT is also given. Finally, the processing steps and its computation issues are analyzed for the Bayesian parametric detectors and their detection performances are evaluated via Monte Carlo simulations. The results show that the parametric matched filter in Bayesian framework is better than GLRT, when the training data are not sufficient.