Optimized Calibration of Terahertz Polarimetric Imaging Systems With Imperfect Polarizers for Accurate Jones-Matrix...

Accurate terahertz (THz) polarimetric imaging requires robust system calibration to recover the Jones matrix of a sample. In many THz systems, calibration procedures often rely on wire-grid polarizers (WGPs) that exhibit frequency-dependent leakage when the wire period is comparable to the operating wavelength, which can introduce systematic errors. In this work, we develop an in situ calibration framework for a polarimetric THz scanner in which an imperfect WGP is rotated in front of a mirror placed at the sample position to provide a complete set of calibration measurements. The calibrating WGP is modeled as a lossless but leaky element having a frequency-dependent extinction coefficient and phase retardance. We derive two equivalent calibration formulations, one of which is based on an orthogonal harmonic-basis formulation that enables a Fourier series interpretation and suggests selecting rotation angles that are uniformly spaced modulo $\pi$ . The condition numbers of the two alternative matrix equations are minimized to optimize the performance of the system with respect to the choice of polarizer rotation angles. Using this approach, calibration can be performed with a minimal set of three optimized angles, yielding a relative error of less than 10% compared to a reference calibration with a full set of 18 angles. We validate the resulting sample characterization procedure using a rotating x-cut lithium niobate crystal, where accounting for WGP leakage removes systematic bias and produces reflection coefficients consistent with the theoretical predictions. Finally, we demonstrate accurate polarimetric imaging by mapping the Jones matrices of two birefringent crystal samples with unknown principal axes via eigendecomposition of calibrated measurement matrix and compared with theoretical values. This optimized calibration technique enables faster and more reliable THz polarimetric imaging in field-deployable scenarios by minimizing the number of required calibration measurements.