Terahertz Imaging of Multi-Level Pseudo-Random Reflectance

This paper introduces a terahertz (THz)-based absolute positioning system with a single THz transceiver as the read head and a multi-level pseudo-random reflectance pattern (e.g., multi-level m-sequences) as the high-resolution scale in a compressed scanning mode. One of key technical challenges here is to computationally recover the multi-level pseudo-random reflectance pattern from compressed measurements. To this end, we develop a variational Bayesian approach to exploit the finite alphabet of reflectance levels and enable a pixel-wise iterative inference for fast recovery. Numerical results confirm the effectiveness of the proposed method.