Ultralow-light hyperspectral imaging via broadband Fabry–Perot resonance and single-photon detection

Hyperspectral imaging in ultralow-light conditions remains challenging due to the low signal-to-noise ratio. To address this challenge, we developed an ultralow-light hyperspectral imaging system that combines broadband Fabry–Perot spectral modulation with single-photon detection. We introduced a spectral–spatial dual-domain attention network to enhance spectral accuracy and provide fine-grained high-resolution information by capturing long-range spectral dependencies and achieving 2× spatial upsampling. To ensure consistency with real physical imaging processes, we simulated a low-light hyperspectral dataset across 28 wavelength channels from 450 to 650 nm at 0.3 lux. Experiments on multiple macroscopic scenes and fluorescent microspheres validate that the proposed method can achieve high spectral fidelity, providing a reliable solution for hyperspectral imaging in low-light conditions.