基于时频域增强和全变差的群目标信号分离

针对低分辨雷达获取的群目标信号的弱时频正交性以及难以分离的问题,在进行时频域增强处理的基础上,提出了一种基于全变差(TV)的群目标信号分离方法。在旋转目标模型的基础上,首先通过分析群目标信号的稀疏性,指出了进行时频域增强处理的必要性。然后利用群目标中各子目标对应的微动周期的差异性,通过双向延迟处理,对多次观测得到的群目标信号进行时频域增强处理。最后根据群目标信号能量区域的分布特性,利用局部TV融合和主分量分析相结合的方法,实现了群目标信号的高保真分离。仿真结果表明,在采样率较低的情况下,文中方法有效地解决了群目标信号中弱信号分量的分离及提取问题,其融合分辨效果明显优于基于TV范数的融合方法。

Group-target signal separation based on time-frequency enhancement and total variation

To solve the problems of weak time-frequency orthogonality and the complexity of separation on group-target echo acquired by low-resolution radar, a group-target signal separation algorithm based on total variation (TV) is proposed as a precondition of time-frequency enhancement. The necessity of time-frequency enhancement is indicated by analyzing the sparsity of group-target signal on the basis of rotating model. According to the differences between the micro-motion periods of different sub-targets, the multi-view group-target echoes are respectively enhanced in time-frequency domain by two-way delay processing. Ultimately, on the basis of the distribution property of energy region for group-target echo, the group-target echo is separated with high fidelity via the regional TV fusion method in conjunction with principal component analysis. The simulation results validate that the algorithm can be used to separate and extract weak signals from some strong signals more easily when the sampling rate is lower. And the fusion resolution of the proposed algorithm is also superior to the fusion method based on TV norm.