基于固有模态奇异值熵的微弱目标检测算法

Weak Target Detection Based on the Entropy of the Intrinsic Mode Singular Value

为检测海杂波中的微弱目标,文中首先提出采用固有模态函数构建海杂波的特征矩阵,解决采用延时嵌入法构建时间序列的特征矩阵时嵌入维数和延迟时间难以确定的困难,然后经奇异值分解获得海杂波的固有模态奇异值,并分析海杂波固有模态奇异值的特点以及目标的影响。分析发现,当目标出现时,海杂波的后6个固有模态奇异值明显增大,而固有模态奇异值熵可以描述目标对海杂波固有模态奇异值的这种影响。在此基础上,文中提出了采用固有模态奇异值熵检测微弱目标的方法。仿真结果表明,与基于盒维数的微弱目标检测算法和多脉冲单元平均恒虚警检测算法相比,该文所提算法的检测性能较好。

To detect the weak target in the sea clutter, the intrinsic mode singular values of the sea clutter are obtained by the singular value decomposition, then the characteristics of its intrinsic mode singular values and the influence of the target on the singular values are analyzed, after the eigen matrix is constructed by the intrinsic mode functions to overcome the difficulties of ascertaining the embedding dimension and delaying time in the eigen matrix construction by the delaying embedding method. It is found that the latter 6 intrinsic mode singular values become larger when the target occurs, that can be described by the entropy of the intrinsic mode singular values, which is proposed to detect the weak target. As shown of the simulation results, the detection performance of the method proposed in the paper is better than the one of the method based on the box dimension and the muhi-pulses CA-CFAR method.