Application Of Wavelet Transform In Terahertz Nondestructive Testing Signal Preprocessing

Terahertz time-domain spectroscopy,as an emerging coherent detection technology,quickly and accurately acquires the internal information of non-polar materials,occupies an important position in the field of non-destructive testing.During the detection process,the signal-to-noise ratio(SNR)of terahertz signal is reduced by several factors,such as absorption of experimental environmental noise and materials and interbedded multiple reflections,which affect the acquisition and interpretation of the subsequent non-destructive testing information.While the traditional Fourier transform has limitations in denoising non-stationary signals such as terahertz signal,wavelet transform owns the advantages of time-frequency localization and multi-resolution analysis in processing signals.Therefore,wavelet transform was applied to the preprocessing of terahertz non-destructive testing signal in this research,and the major research contents were as followings:(1)Reflective air terahertz time-domain signal was taken as the experimental object to study the selection process of the optimal wavelet denoising combination.Given the characteristics of terahertz time-domain signals,?-? evaluation rule was creatively proposed for evaluating the denoising effect of all parameters in the process of wavelet denoising that overcame the disadvantage of the large deviation between the theoretical optimal solution and the actual situation in the traditional evaluation rules of SNR and root-mean-square error.The optimal solutions for denoising parameters of wavelet basis,decomposition level and threshold rule were selected based on the ?-? evaluation rule.The results showed that the ideal terahertz signal denoising results could be obtained by using the combination of Sym7 wavelet,decomposition scale = 5,minimaxi threshold selection scheme,soft threshold method,and Mln reset method.It has been verified by a large number of experiments that the conclusion can be widely applied to the majority of terahertz time-domain signals.(2)Wavelet transform was applied to the optical parameter extraction of materials.An optical parameter extraction model was established,a transmission one-point testing experimental platform was built,and software programs were developed,so as to obtain and compare the optical parameters of the polycarbonate samples acquired before and after wavelet denoising pre-processing.The experimental results indicated that the standard deviation of refractive index was 0.0043 after wavelet denoising pre-processing that was superior to that before the pre-processing which was 0.2177;the average refractive index was 1.657 with a higher accuracy compared to the pre-processing;the oscillation of absorption coefficient spectrum was eliminated significantly,and the characteristic absorption peak that was not recognized before the pre-processing was extracted at 0.72 THz.(3)wavelet transform was applied to the detection of void defect inside materials.A short-term integral tomographic method was innovatively proposed,which is more suitable for characterizing the section information inside materials than the traditional methods.A reflective non-destructive testing imaging experimental platform was built and software programs were developed,in order to obtain the testing images of three defective samples: high density polyethylene,phenolic plastic and glass-fiber reinforced plastic,before and after wavelet denoising pre-processing.According to the image characteristics,it was suggested to employ Weber contrast model and defect recognition rate to comprehensively evaluate the imaging effect.The experimental results showed that wavelet denoising pre-processing significantly improved the imaging quality problems such as poor contrast effect and noise band interference caused by material absorption and interbedded multiple reflections,thereby to effectively enhance the defect detecting capability.In this paper,the wavelet transform has been used to improve the quality of the initial signal data,optimizing the results of terahertz nondestructive testing.Other than the existing denoising means that focuses on the processing of the subsequent detection results,this method provides a new solution for extracting the real information of materials under noise background.