High-fidelity single-photon full-color imaging via frequency-spectrum cross-channel correlation

Single-photon color imaging has great application potential in biomedical imaging, environmental monitoring, and security surveillance. However, the inherent photon randomness and the neglect of channel correlation in single-photon imaging restrict its applications. We propose a high-fidelity single-photon full-color imaging technology designed for photon-starved conditions. Through cross-channel correlation technology, the color channels extracted from frequency-spectrum harmonics are fused with the photon-counting channel, enabling effective noise suppression, complementary information integration, and color ratio correction. Experimental results show that the proposed technology can realize full-color imaging with distinctions of over 4500 colors per pixel. While achieving an ultrabroadband color gamut, the technology maintains the overall color deviation below the threshold of human visual perception, enabling high-fidelity full-color reconstruction. Moreover, the randomness inherent in photon statistics is exploited to enable frequency-domain compressed sensing, which markedly reduces the number of photons required for demodulation. The technology can obtain dynamic imaging with 32 frames per second, even with the reception of only one photon per pixel. These results highlight the application potential of this technology in full-color imaging under photon-starved conditions.