An Adaptive Threshold Shearlet Filtering Function For GPR Voids Data Preprocessing

With the acceleration of urban construction,road,railway and tunnel traffic has developed rapidly.Under a large number of construction or environmental impacts,the problem of underground voids or tunnel lining voids has become increasingly severe,resulting in potential traffic safety hazards.Ground Penetrating Radar(GPR)is an effective non-destructive testing device for detecting underground or through-wall voids.When using GPR to detect road and tunnel lining void distress,the void reflection signal in the received data is often interfered by random noise and coherent noise,and the direct wave is the main component of coherent noise.Because the conventional GPR data preprocessing method can not effectively meet the high precision void detection.For the problem that the signal to noise ratio in the GPR void data is low and the void reflection signal is interfered by the direct wave.In order to solve this problem,this thesis carries out the following work:(1)The random noise and direct wave suppression methods of GPR void reflection signals are studied.Because of the superiority of Shearlet transform method on the denoising of seismic signals,the seismic echo and the GPR echo have certain similarities,this thesis intends to use the Shearlet transform method based on threshold function filtering to process GPR void data.However,traditional Shearlet transform methods often use fixed threshold function filtering,which leads to effective signal and energy loss.In order to improve the effect of Shearlet transform on the GPR void data preprocessing,this thesis starts from the nature of Shearlet transform,multi-scale and multi-directional decomposition of GPR signals,and constructs adaptive threshold function based on the distribution difference of effective signal and noise in Shearlet domain.And obtain a threshold for each unprocessed shearlet coefficient according to an adaptive method,improving the shortcomings of the traditional hard threshold function discontinuity,and use this function to adjust the Shearlet coefficient to achieve adaptive filtering.For the direct wave signal,based on the distribution difference between the direct wave and the effective signal coefficient in the Shearlet domain,the Shearlet coefficient of the direct wave is separated,and suppressing direct waves by adjusting the Shearlet coefficients associated with direct waves,thereby achieving the purpose of suppressing the direct wave.(2)Based on the Finite Difference Time Domain(FDTD)method,the GPR echo signals of road void and tunnel lining void are simulated.The adaptive threshold Shearlet transform method constructed in(1)is used to suppress the noise and direct wave of the forward modeling void data and the field data for processing.The processed A-scan signal and B-scan data show that the random noise and direct wave components are effectively suppressed.The road and tunnel void data of the forward simulation are compared with the median filtering,wavelet transform and curvelet transform denoising methods respectively,the results show that the adaptive threshold Shearlet transform method improves the signal to noise ratio of road and tunnel lining void data by 13 dB and 12 dB respectively.Finally,the mean filtering method and principal component analysis method are used to compare the effect of GPR field data to remove direct wave.The results show that the Shearlet transform method is relatively optimal in improving the signal to clutter ratio and the image clarity of the target image,which verifies the effectiveness of the adaptive threshold Shearlet transform method in the preprocessing of GPR void detection data.