Dielectric properties characterization: A simple inverse problem approach

Thickness measurement by model-based exhaustive analysis in far-infrared

“…Nevertheless, the broad range of frequencies used by TDS and FMCW systems requires high sweep acquisition times compared to fixed-frequency imaging. Therefore, simply using several frequencies could be a trade-off solution to generate accurate results in reasonable time, as implemented in the genetic and evolutionary material parameter reconstruction field [5,6,7,8]. In these latter studies, the authors solved the inverse problem by giving hints about the parameters or using adaptative learning algorithms.…”

“…However, real time assessment is hampered by the use of genetic algorithms which are known to converge slowly and require tuning of multiple hyperparameters. The research of Zwick et al has been followed by studies by numerous authors using genetic or evolutionary algorithms and artificial neural networks to retrieve material parameters [6,7,8]. In addition to the fact that these methods cannot be used for real-time imaging, they also dedicated to single-location assessment, not to complete MLS assessment.…”