一种信号检测预处理的改进多尺度形态学滤波方法  

An Improved Multi-scale Morphological Filtering Method for Signal Detection Preprocessing

作　　者：宁明峰 NING Mingfeng(Southwest China Institute of Electronic Technology,Chengdu 610036,China)

机构地区：[1]中国西南电子技术研究所,成都610036

出　　处：《电讯技术》2023年第4期499-504,共6页Telecommunication Engineering

摘　　要：通信信道冲击噪声干扰和宽带接收机非线性影响会导致接收的信号噪声基底不平坦,若不考虑噪声基底起伏而直接进行检测信号,可能会导致漏检。为提高信号检测精度,需估计信号噪声基底。为解决该问题,提出了一种信号检测预处理的改进多尺度形态学滤波方法。该方法采用多尺度开-闭组合运算对信号频谱进行形态学滤波,利用不同尺度的结构元素对不同带宽的信号进行局部滤波,并利用滤波后凸包的存在性作为滤波终止条件。仿真结果表明,该方法可以有效将不同带宽的信号滤除,有效提取噪声基底噪声,同时可消除频谱中低电平毛刺对噪声估计的影响。The influence of the impact noise of communication channel and the nonlinear of the wideband receiver will lead to uneven signal noise.If the signal is detected directly without considering the fluctuation of uneven noise,it may lead to missed detection.To improve the accuracy of signal detection,it is necessary to estimate the signal noise before signal detection.A pretreatment method using multi-scale morphological filter is proposed to solve this problem.This method uses the multi-scale open-close combined operation to filter the signal spectrum,adopts structural element of different scales to locally filter,and takes existence of the filtered convex hull as the filtering termination condition.The simulation results show that the proposed method can effectively filter signals with different bandwidths,extract the noise effectively,and eliminate the influence of low level burrs in the spectrum on noise estimation.

关 键 词：信号检测 多尺度形态学滤波 噪底估计 结构元素 

分 类 号：TN911.72[电子电信—通信与信息系统]