一种基于KL分离度的改进矩阵CFAR检测方法

摘要

矩阵 CFAR 检测器是根据信息几何理论提出的，但其恒虚警特性没有从理论上得到分析，且检测性能也有待进一步提高。该文首先根据矩阵流形上正态律的概念从理论上推导了矩阵 CFAR 检测器的恒虚警特性，并在此基础上，利用积累性能更好的KLD(KULLBACK-LEIBLER Divergence)代替测地线距离，提出了一种改进的矩阵CFAR检测器。最后通过仿真实验验证了改进方法具有更好的检测性能。%The matrix CFAR detector is proposed according to information geometry theory, but its constant false alarm property is not analysed, and the matrix CFAR’s detection performance still needs to be improved. Firstly, the matrix CFAR’s constant false alarm property is analysed according to the normal law on matrix manifold, on this basis an improved matrix CFAR detector is proposed with replacing the geodesic distance with KULLBACK-LEIBLER Divergence (KLD). Finally, simulation experiments verify that the improved method has better detection performance.