Terahertz diffractive imaging with saturated data inpainting

Abstract

Multiplane iterative phase retrieval is a promising approach to diffraction imaging, which accurately determines the topographic and internal characteristics of various objects. Nevertheless, the detection systems used often have a limited dynamic range, resulting in overexposure of recorded intensity distributions. In this Letter, we present a novel, to the best of our knowledge, reconstruction algorithm that inpaints saturated areas on the measured intensity datasets and reliably retrieves wave complex amplitude. The proposed technique can be used in various spectral ranges, while we have tested it in the terahertz frequency range, where the problem of sources and detectors is most acute. We show that retrieved amplitude and phase distributions have a quality comparable to that of the images reconstructed from the reference high dynamic range technique. Herewith, the proposed approach seriously simplifies the process of data acquisition, what expands the possibilities in the design of measurement tools and studies of dynamic scenes.