Implementing Fast Terahertz Imaging through Sparse Algorithm

Terahertz wave imaging, renowned for its non-contact and non-destructive attributes, has found widespread application across various industrial detection domains. However, its sluggish detection speed has hindered further deployment in numerous industrial scenarios. This paper introduces a terahertz fast imaging algorithm by integrating sparse sampling and predictive reconstruction techniques, facilitating rapid imaging of micron-scale arc-shaped air gap defects commonly encountered in XLPE insulation. Results demonstrate that compared to conventional imaging methods, the proposed approach can enhance imaging speed by an order of magnitude while preserving detailed imaging features. Moreover, through an examination of the correlation between sparse coefficients and imaging quality, a sparse threshold determining the retention of imaging detail information is identified. Finally, by leveraging the Gaussian beam properties of terahertz waves, the physical principles underpinning the sparse threshold are explored, culminating in a mathematical expression thereof. This work presents a novel avenue for accelerating terahertz imaging and derives threshold conditions for sparse imaging, thereby furnishing both empirical evidence and theoretical formulations to aid the industrial sector in compressing terahertz imaging data and enhancing imaging speed.