The Application Of Wavelet Transform For GPR Signal

As a novel highway nondestructive testing technology with the advantages of fast, continual process, flexible operation, high-resolution and nondestructive, Ground Penetrating Radar (GPR) has be widely applied in traffic construction domain.The signal used by GPR always includes many peaks or mutational non-stationary components. And the pretreatment is need before signal analysis to remove the noise part and extract the useful signal. Compare with the traditional Fourier transform's uselessness, the wavelet transform is a more suitable mean in GPR signal process, because it can distinguish the mutation and noise effectively. In the discrimination of roadbed and tunnel lining quality by wavelet transformed GPR image, the core content of GPR image explanation is identify the object's radar image character correctly.In this dissertation, the constitution of GPR and its testing principle and the basic theory of wavelet analysis are introduced at first. Compare with the advantages and disadvantages of wavelet transform's traditional threshold function, a new threshold function denoise algorithm is presented for single-wave radar signal based on the character of radar signal, and a mean of void-containing layer property recognition is given based on the application of wavelet transform. Then aim to the irrelevance between signal and noise of GPR at the same time, we present a better filter mean (united denoising algorithm) which combined wavelet packet, KL transform and median filter. Finally, the validity and denoise effect of the mean are demonstrated by simulation using a serial of measured data in this dissertation.