Low-dimensional Composite Grating Array Broadband Filter for Visible-Nir Spectrum Reconstruction

A micro-nano structure spectral filter facilitates bandpass imaging with a large optical flux and strong spectral curve controllability, which reduces the difficulty of processing and complexity. Therefore, considering the thin film transport theory and the finite element method, a visible-near infrared (400-900 nm) wide-band spectral filter satisfying the target spectral reconstruction requirements was fabricated. Employing lithography, coating, and lift-off processes, 24 groups of composite grating array broadband filter film lines were fabricated on the same substrate. The filter and optical microscope were packaged at the front of the GSENSE5130 detector of a large surface array in a dust-free environment to ensure the parallel placement and utilization rate of the membrane system and detector image element. Finally, a three-segment log-regularization constrained convex optimization algorithm (IReg-Cvx algorithm) based on a low-dimensional spectral curve was employed to spectrally reconstruct the multi-group ground target spectral curve. The detector errors in the gray values of the same set of transmission curves were +/- 2%. The proposed IReg-Cvx algorithm was confirmed to outperform the standard regularization algorithm in terms of mean square error and relative reconstruction errors. Moreover, the spectral reconstruction accuracy was higher upon the application of the IReg-Cvx algorithm and the target spectral features were stably captured. The low-dimensional compound grating array broadband filter facilitated high-precision spectral reconstruction of the visible-near infrared spectrum and is applicable for object-oriented data acquisition for identifying ground objects and targets. Therefore, it can aid in the design and development of compact and lightweight star-borne imaging spectrometers.