基于伪逆算子的超分辨率信号处理方法研究

Research of super-resolution signal processing method based on pseudo inverse operator

为了解决信号压缩过程中旁瓣串扰问题并提高参数估计时频率的估计精度,在研究超分辨信号处理理论的基础上,从框架理论入手,提出了可改善系统性能的恒噪声灵敏度分解算法。首先在基于矩阵表示的超分辨信号处理模型的基础上,采用矩阵伪逆算子,进行基于信号压缩的微波成像处理,讨论了伪逆算子应用于信号分解过程中的噪声性能;然后针对伪逆算子导致系统噪声增益增加的问题,通过根据信号处理的噪声指标要求设定处理门限,实现对信号的超分辨处理。仿真实验结果表明,该方法可消除匹配滤波类算法不可避免的旁瓣串扰问题,并可在一定程度上改善系统分辨率性能。

To resolve the side-lobe coupling problem during signal compression, and improve estimation accuracy of frequency during parameter estimation, this paper based on super-resolution technique of signal processing, in the viewpoint of the frame theory, proposed a constant noise gain decomposition method. First, based on the matrix expression of the super-resolution signal model, using the pseudo inverse operator, the microwave imaging processing via signal compression technique is performed, and the noise performance of pseudo inverse operator is discussed. Then, to reduce the noise gain of pseudo inverse operator, a threshold is added in the pseudo inverse according to the expected noise gain, and obtained super-resolution processing result eventually. The simulation experiments show that the pseudo inverse operator and constant noise gain decomposition method can reduce the side-lobe coupling of matched filtered method, and improve the resolution of the system to some extent.