| 摘要: |
| 针对大擦地角复杂海况下雷达目标检测中虚警率难以精确控制的问题，提出一种基于非参数边缘估计与Copula理论的多特征虚警率控制检测方法。首先提取相对平均幅度和相对多普勒幅度峰值两类特征，利用核密度估计自适应拟合杂波特征的经验累积分布，通过概率积分变换将特征映射至均匀空间。针对实测数据呈现的对称尾部依赖特性，选用Frank Copula刻画特征间相依结构，基于Copula-CFAR定理推导给定虚警率下的解析门限。利用X波段雷达实测数据进行验证，并与CA-CFAR、三特征检测器、时频三特征检测器、Lin-DBSCAN-CFAR及Copula-CFAR对比。结果表明:所提方法的检测性能优于所有对比方法，验证了非参数建模在提升虚警控制精度方面的有效性。该方法为复杂海况下多特征虚警率控制检测提供了新的理论支撑和技术途径。 |
| 关键词: 雷达目标检测 海杂波 Copula理论 核密度估计 |
| DOI： |
| 分类号:TN959 |
| 基金项目:国家自然科学基金项目（面上项目，重点项目，重大项目） |
|
| Copula-Based Target Detection Method with False Alarm Control and Performance Analysis Under High Grazing Angles |
|
|
|
|
| Abstract: |
| Aiming at the problem that the false alarm rate in radar target detection is difficult to control precisely under complex sea states with high grazing angles, a multi-feature false alarm rate control detection method based on non-parametric marginal estimation and Copula theory is proposed. First, two types of features, namely relative average amplitude and relative Doppler amplitude peak, are extracted. Kernel density estimation is utilized to adaptively fit the empirical cumulative distributions of the clutter features, and the features are mapped into a uniform space via the probability integral transform. Considering the symmetric tail dependence characteristic exhibited by the measured data, the Frank Copula is selected to characterize the dependence structure between the features, and the analytical threshold for a given false alarm rate is derived based on the Copula-CFAR theorem. The proposed method is validated using measured X-band radar data and compared with CA-CFAR, the tri-feature detector, the time-frequency tri-feature detector, Lin-DBSCAN-CFAR, and Copula-CFAR. The results show that the detection performance of the proposed method outperforms all compared methods, verifying the effectiveness of non-parametric modeling in improving the accuracy of false alarm control. This method provides new theoretical support and a technical approach for multi-feature false alarm rate control detection under complex sea states. |
| Key words: radar target detection sea clutter copula theory kernel density estimation |