基于同步挤压S变换的探地雷达信号消噪方法

De-noising Method of Ground Penetrating Radar Signal Based on Synchrosqueezing S-transform

为提高对探地雷达低幅度有用信号的识别效果,提出了基于同步挤压S变换的消噪方法.首先将含噪声信号进行S变换,并计算出信号的瞬时频率,再将信号的时频能量进行重新集中,从而获得高分辨率的压缩时频谱.然后根据噪声出现的时间和频率,设计相应的二维滤波器在时频域对信号进行消噪.在消除噪声能量后,再进行反变换得到有用信号.由于在压缩时频谱上可以更好地区分有用信号和噪声,消噪时有效信号的能量可以被尽可能多地保留下来.仿真信号和探地雷达实际信号的处理结果表明,利用同步挤压S变换在时频域进行消噪能够取得很好的效果.

In order to better identify useful low-amplitude ground penetrating radar(GPR)signals,a de-noising method based on the synchrosqueezing S-transform(SSST)is proposed in this paper.Firstly,the new method performs S-transform onto a signal containing noise,calculates instantaneous frequencies of the signal,and relocates the energy of the signal to vicinity of the instantaneous frequencies to obtain high resolution compressed time-frequency spectrum.Then,according to the time and frequency coordinates of the noise,two-dimensional filters are designed to remove noise from the signal in the T-F domain.After the noise being removed,the useful signal can be restored after performing inverse transform on the T-F spectrum.Since the energy of the useful signal and the noise can be more clearly distinguished on the compressed T-F spectrum,the energy of the useful signal would be better retained.The processing results of a simulated signal and GPR data show that the T-F domain de-noising method using SSST can obtain better results than using the conventional T-F transforms.