Journal
JOURNAL OF INFRARED MILLIMETER AND TERAHERTZ WAVES
Volume 42, Issue 9-10, Pages 929-946Publisher
SPRINGER
DOI: 10.1007/s10762-021-00819-1
Keywords
Nondestructive testing; Terahertz imaging; Sparse deconvolution; Preconditioned conjugate gradient; Dispersion
Funding
- Conseil Regional Grand Est
- CPER SusChemProc
Ask authors/readers for more resources
Terahertz time-of-flight tomography (TOFT) faces challenges in characterizing samples with both micron-scale and millimeter-scale layers, due to obscured echoes and mistaken overlapping echoes from thin layers. A sparse-deconvolution (SD) technique demonstrated in this paper overcomes these challenges, successfully extracting impulse responses of samples with thick and thin layers, providing a clearer and more accurate reconstruction.
Terahertz (THz) time-of-flight tomography (TOFT), a nondestructive-evaluation technique for the stratigraphic characterization of structures with layers on the micron-to-millimeter scales, has proven to be challenging to apply to samples containing both micron-scale and millimeter-scale layers. In THz TOFT, echoes reflected from distant interfaces and defects are often obscured as they may be immersed in a noisy background as such features in the reflected signal may be weak due to attenuation and dispersion, leading to the loss of valuable information. Moreover, overlapping echoes from any optically thin layers, such as thin coatings on thick specimens, are likely to be mistaken for a single interface in reconstructing the stratigraphy. Thus, layered structures containing both thick and thin layers have proven problematic for THz TOFT characterization. In this paper, a sparse-deconvolution (SD) technique, based on an interior-point method, and including a propagation model accounting for dispersion is demonstrated. The method is shown to be successful in extracting the impulse response of samples that combine the challenges of both thick and thin layers. The robustness and effectiveness of this method are verified numerically and experimentally. While the proposed SD approach does not perform as well as cross-correlation (CC) techniques in terms of the maximum thickness, it can provide a clearer and more accurate reconstruction of moderately thick samples incorporating thin layers.
Authors
I am an author on this paper
Click your name to claim this paper and add it to your profile.
Reviews
Recommended
No Data Available