Gpr Signal Denoising Method Based On Wavelet Transform

Ground penetrating radar (GPR) is an effective tool to detect buried targets, which detects buried targets via the discontinuity of dielectric. Ground penetrating radar has a lot of virtues, such as rapid and continuous surveying, simple and convenient operation, high resolution, nondestructive detection etc, so which has been used to fields of national defence, police, construction of highway, railway, bridge, tunnel, and prospecting, geologic studies, archaeological studies etc.The GPR signaltaken by the rush often include many or mutation of the non-stationary components, and the noise is not necessarily a smooth white noise. To analyze these signals, the first signal of the need for pretreatment of signal noise removal of part, extract useful signal. This right signal denoising, using traditional Fourier law is powerless. Wavelet transform is an analyisis method for non-stationary signal in recent years, and it is considered as the breakthrough of Fourier analysis. It also get over the disadvantage of local weak time-frequency analysis existed in Fourier analysis, what's more, it can analyze signal in different time scales and frequency scales. It is obviously that wavelet analysis can distinguish signals of the mutation and noise, thus achieving the signal denoising. So in ground penetrating radar signal processing, wavelet transform is more suitable.Three wavelet transform de-noising methods are studied in this paper on the basic of wavelet analysis theories. The processing results indicate that coefficient of high frequency place to zero method has the more advantages than traditional filter, although it also have some limitations, and this method is more conveniently; the threshold de-noising method has widespread applicability and it is gives multi-thresholds in this paper,specially giving the new threshold function aim at the existent disadvantages in the traditional soft threshold and hard threshold,at the same time it is compare the advantages and disadvantages of each threshold.Its effect of de-noising is very good through simulation research, specially the new threshold function de-noising methold; Besides the Mold maximum value wavelet transform de-noising has advantage in the low signal-to-noise ratio date, and the method fits in the condition that contains white noise and many singularities.The paper applies the three methods upwards to the GPR data processing in the end, proves the validity of these methods through the practical GPR data, then achieves the goal of de-noise the signal in GPR data processing.