Improved algorithm for material characterization by terahertz reflection imaging

We propose a new algorithm to improve the accuracy of sample characterization when the sample is measured on the imaging window of a terahertz reflection imaging system. Most existing approaches assume that the imaging window is homogenous, but we have noticed that there are small variations in thickness across such windows and that these can significantly affect the accuracy of standard approaches to extract the sample properties, particularly the absorption coefficient. Our algorithm accounts for both thickness variation across the imaging window as well as fluctuations in the incident pulse and mechanical jitter. Furthermore, our algorithm removes the need to measure the reference at every point that the sample is measured and thus reduces the imaging time.