Terahertz Non-destructive Evaluation of Layered Media with the Maximum Likelihood Estimator

Terahertz (THz) non-destructive evaluation (NDE) has shown great promise for applications in manufacturing, security screening, and medical imaging. However, interpretation of the THz A-scans is complicated by a number of factors. Other researchers have applied signal processing techniques that improve object boundary detection, but still rely on human interpretation without a statistical framework that provides automated detection. This work presents a statistical signal processing algorithm to detect boundaries/defects in THz NDE data based on techniques that have traditionally been used to detect targets in radar/sonar applications. It is demonstrated that this method provides better signal-to-noise ratio (SNR) in the processed waveform along with a mature mathematical framework for analysis of measured data.