Accurate reconstruction of terahertz spectral images with enhanced spatial resolution via complex mapping.

The wavelength of terahertz waves varies by two orders of magnitude. Long-wavelength terahertz images suffer from low spatial resolution due to the millimeter-level diffraction limit. Conventional resolution-enhancing methods are generally limited by sample types and field of view. To overcome these challenges, we propose a resolution-enhancing algorithm for terahertz spectral imaging. This algorithm leverages the advantage of ultra-broadband complex spectral imaging and determines the mapping relationship between the short- and long-wavelength images through clustering and genetic algorithm optimization. The numerical modality supports nearly all optical configurations and sample types. Transmission and reflection measurements validate the superior performance, demonstrating up to 6-fold resolution improvement. Moreover, the complex spectra can be accurately recovered, enabling precise extraction of broadband complex permittivity and subsequent analysis for sub-diffraction-limit objects.