RESEARCH ON WAVELET EXTRACTION OF GPR SIGNALS BASED ON MULTILEVEL FRACTIONAL FOURIER TRANSFORM FILTER

摘要

The traditional ground penetrating radar (GPR) convolution model is discussed in this paper. It proposes the use of linear time-varying systems to describe radar echo signals in which the radar echo signal is described to time integral form between the wavelet and formation. The model is used in the process of extracting a GPR wavelet using Fractional Fourier transforms (FrFT). A single optimum filter is insufficient when the signal is affected by random noise and the formation is a random medium; thus, a concept of multilevel optimum filtering is proposed. A multilevel FrFT filter comprises single-stage FrFT filters in series connection. First, the input signal is transformed before it enters the domain where it is multiplied by the filter function. After an inverse transform, it enters the subsequent region through a transform. This process is repeated many times until optimum filtering is achieved and the effects of noise are eliminated. Then, the wavelet of the GPR synthetic and real data is extracted using an algorithm. The obtained results show that the wavelet extraction algorithm is valid, verifying that the proposed model is reasonable and effective.