GPR noise reduction based on wavelet thresholdings

Ground-penetrating radar (GPR) is a nondestructive tool, and its data contain information about media beneath the earth's surface. Because of the large bandwidth of the impulse GPR system, the GPR signal is contaminated by the noises coming from various sources. De-noising the GPR signals before performing any data analysis is very important in order to enhance the detection performance of the GPR and to allow accurate layer depth prediction. This paper focuses on denoising the GPR data by various thresholding rules such as hard, soft, firm shrinkage, and nonnegative garrote thresholding rules. We have decomposed the signal by dual-tree complex wavelet transform(DTCWT) to some coefficients and the shrink ones by various shrinkage rules based on different thresholding functions. For comparison between the results we supply a table which contains the signal-to-noise ratio (SNR) and root-mean-square error (RMSE) of all methods which have been used.