Fig. 2
From: Deep learning-based single-shot computational spectrometer using multilayer thin films
DL-based single-shot computational spectrometer. (a) Schematic of the experimental setup. A collimated beam from a light source is split into two beams by a beam splitter (BS) after passing through a color filter and a long pass filter (LPF). The spectrum of a beam is measured using a commercial spectrometer. The other beam is modulated by the MTF filter array and captured as an image by a CMOS camera. (b) DL architecture comprises a dense layer and a U-Net backbone architecture with residual connections. (c) Examples of recovered test spectra using the trained DL architecture. Dashed black lines represent GT spectra measured using a commercial spectrometer. Solid blue lines represent reconstructed spectra using the proposed spectrometer.
