HOMOMORPHIC SIGNAL PROCESSING ALGORITHM OF GROUND PENETRATION RADAR

Abstract

The actual problem of measuring the thickness of the surface layers of the soil using GPR radar systems under stringent restrictions on the minimum length of a video pulse and under the presence of the noise caused by resonance effects in antenna and multiple reflections of the sounding signal from the boundaries between media has been studied. Traditional extreme correlation signal processing methods do not provide the resolution less than half of the total duration of the pulse. To overcome this limitation is possible by using nonlinear methods, for example the homomorphic signal processing. The possibility of high resolution of subsurface stratified media boundaries by using non-linear homomorphic digital signal processing of reflections is proved theoretically and experimentally. The proposed method has used the non-linear spectrum transformation of ultra wideband sounding signals. It allows getting resolution approximately to 5 % of pulse duration instead of limitation of the traditional correlation method of signal processing. The FFT algorithm is used to calculate the amplitude spectrum of the signal, the result is subjected to logarithms and then the inverse Fourier transform is computed, to get a signal's cepstrum containing the probe signal and digital signal delay. The analysis of cepstrum at a relatively high signal to noise ratio allows one to calculate the estimate of the thickness of the layer. The results of simulations and experimental data processing are also shown.

Keywords

• amplitude spectrum
• boundary between the media
• cepstrum
• fast fourier transformation
• ground penetrating radar
• identification
• pulse
• spectrum
• ultra-wideband signal