Fig. 1: Framework of the computational spectrometer with Single-Spinning Film Encoder (SSFE) and deep learning. | Communications Engineering

Fig. 1: Framework of the computational spectrometer with Single-Spinning Film Encoder (SSFE) and deep learning.

From: A computational spectrometer for the visible, near, and mid-infrared enabled by a single-spinning film encoder

Fig. 1

a The optical setup of the system, with a broadband light source, sample, and the SSFE-based spectrometer, including a polarizer, SSFE, and detector arranged sequentially. b Structure of the 10-layer SSFE composed of alternating high-index TiO₂ and low-index SiO₂ on a sapphire substrate. c Diagram illustrating the rotation of SSFE, with a rotational angle ranging from 0° to a maximum value of 70°. Specifically, the rotation angles for P-polarized light are 0°, 10°, 20°, 25°, 30°, 35°, 40°, 55°, 50°, 55°, 60°, 65°, and 70°, while the rotation angles for S-polarized light are 10°, 20°, 25°, 30°, 35°, 40°, 55°, 50°, 55°, 60°, 65°, and 70°. d Workflow of the spectral encoding and reconstruction procedure. Herein, spectral encoding is accomplished by SSFE, and the sensing matrix R is composed of the transmittance of SSFE under different polarizations and spinning angles. Spectral reconstruction is achieved using a deep learning-based reconstruction network featuring fully-connected layers, where the input layer corresponds to the measured light intensities Ii, the output layer represents the reconstructed spectra, and the three hidden layers are employed to capture the complex spectral relationships and enable accurate reconstruction.

Back to article page