基于稀疏表示的太赫兹时域光谱分层检测算法

Terahertz Time-Domain Spectral Hierarchical Detection Algorithm Based on Sparse Representation

太赫兹无损检测是一种非侵入、非接触性的新型检测技术,在无损检测领域极具潜力.但在实际探测中,对于复杂的太赫兹回波信号,如色散回波和重叠回波,仅利用传统的信号处理方法(如直接法、反卷积法)难以满足对飞行时间定位精度的要求.稀疏表示法作为一种太赫兹信号处理的新方法,具有良好的定位精度和抗噪性.提出基于最小绝对收缩和选择算子(LASSO)的稀疏表示法,从复杂的太赫兹回波信号中重建脉冲响应函数,并构建双过完备字典完成对太赫兹反射信号的色散补偿.针对重建脉冲响应函数幅值不准确的问题,提出幅值衰减系数,修正脉冲响应函数幅值,有效提升时域峰-峰值的成像质量.通过数值计算和实验分析,验证了所提方法的有效性.可以预期,基于LASSO的稀疏表示法能够为太赫兹无损检测中的信号处理提供新的解决思路.

Terahertz nondestructive testing(THz-NDT)technology is a new noninvasive and noncontact detection technology with strong penetration capability for nonmetallic and nonpolar composite materials,and has a great potential in the field of NDT.In practical detection,for complex terahertz echo signals,such as dispersive echoes and overlapping echoes,it is difficult to meet the requirements of localization accuracy for the time-of-flight only via traditional signal processing methods,such as the direct and deconvolution methods.In this context,the sparse representation method,as a new method for THz signal processing,has good localization accuracy and noise immunity.In this study,we propose a sparse representation method based on the least absolute shrinkage and selection operator(LASSO)to reconstruct an impulse response function from the complex THz echo signal and construct a double-over-complete dictionary to complete the dispersion compensation for the THz reflection signal.An amplitude decay coefficient is proposed to address the problem of inaccurate amplitude of the reconstructed impulse response function.This correction can effectively improve the peak-to-peak imaging quality in the time domain.The effectiveness of the proposed method is verified through numerical calculations and experimental analysis.The proposed method is expected to provide a novel solution for signal processing in THz-NDT.