Random noise de-noising and direct wave eliminating based on SVD method for ground penetrating radar signals

- Use singular value decomposition to eliminate the random noise and direct wave from ground penetrating radar (GPR) signals
- Compare SVD de-noising method with wavelet threshold de-noising method and bandpass filtering method
- Compare SVD method with the mean trace deleting method in eliminating direct wave
- Give the quantitative criteria on choosing singular values
- SVD method can eliminate the random noise and direct wave in the GPR data validly and efficiently.

In this paper, we present a method using singular value decomposition (SVD) which aims at eliminating the random noise and direct wave from ground penetrating radar (GPR) signals. To demonstrate the validity and high efficiency of the SVD method in eliminating random noise, we compare the SVD de-noising method with wavelet threshold de-noising method and bandpass filtering method on both noisy synthetic data and field data. After that, we compare the SVD method with the mean trace deleting in eliminating direct wave on synthetic data and field data. We set general and quantitative criteria on choosing singular values to carry out the random noise de-noising and direct wave eliminating process. We find that by choosing appropriate singular values, SVD method can eliminate the random noise and direct wave in the GPR data validly and efficiently to improve the signal-to-noise ratio (SNR) of the GPR profiles and make effective reflection signals clearer.