Computational resolution enhancement for dispersive spectrometers based on GMC spectral reconstruction

The spectral resolution of conventional dispersive spectrometers is limited by the combined effects of optical broadening, instrumental response, and detector sampling, making high-resolution spectral measurement costly and technically demanding. Here, we present a generalized minimax-concave-based spectral reconstruction strategy (GMC-SR) to enhance the effective spectral resolution of conventional grating spectrometers without hardware modification. By integrating a nonconvex penalty into a deconvolution framework, GMC-SR suppresses noise and reconstruction artifacts while preserving intrinsic spectral structures, enabling the recovery of spectral features beyond the nominal Rayleigh criterion of the instrument. In simulations, the full width at half maximum (FWHM) for single-wavelength inputs narrows from 0.35 nm to 0.095 nm, and the minimum resolvable spacing in dual-wavelength tests improves from 0.35 nm to 0.20 nm according to an objective peak-to-valley contrast criterion. For broadband interferometric signals, the reconstruction root-mean-square error (RMSE) drops from 0.26 to 0.009, indicating the potential of the method for continuous-spectrum reconstruction. Experiments with a compact, self-built spectrometer further corroborate these gains: the single-wavelength FWHM decreases from 0.28 nm to 0.14 nm, dual-wavelength spacing improves from 0.28 nm to 0.189 nm under the same objective separability criterion, and thin-film measurements achieve an RMSE of 1.0343e-3 with a thickness standard deviation below 0.01 μm. Compared with representative deconvolution and regularization methods, including Gold, Richardson-Lucy, Tikhonov, and total-variation regularization, GMC-SR provides higher reconstruction fidelity with competitive computational efficiency under the tested conditions. These results demonstrate that GMC-SR offers a general, low-cost, and high-fidelity computational route for improving the effective resolution of dispersive spectrometers and for reconstructing both discrete and broadband spectral features across a wide spectral range.