Mid‐Infrared Single‐Photon Computational Temporal Ghost Imaging

Abstract The capture of transient optical waveforms is critical to reveal dynamical phenomena in various fields. However, fast and sensitive mid‐infrared (MIR) measurements are typically limited by processing bandwidth and detection sensitivity of conventional infrared detectors. Here, a computational temporal ghost imaging system is proposed and implemented, which favors high‐speed and high‐sensitivity characterization of MIR temporal objects. The core process relies on high‐fidelity nonlinear optical transduction for facilitating both the programmable structured illumination and frequency upconversion detection based on the high‐performance near‐infrared light modulator and detector, respectively. Consequently, the correlation between the recorded integral upconversion intensity and the designated encoding patterns allows one to reconstruct the MIR profiles with a temporal resolution of 80 ps, well beyond the intrinsic bandwidth or timing jitter of the involved detectors. Moreover, a high detection sensitivity is manifested by recovering single‐photon MIR waveforms with an incident flux below 0.1 photons/bit. Additionally, faithful reconstructions at sub‐Nyquist sampling rates are demonstrated using the compressive sensing algorithm, which can reduce the data acquisition time by over 90%. The presented paradigm features high timing precision, single‐photon sensitivity, and efficient data sampling, which can be extended into far‐infrared or terahertz regions to address pressing demands in fast and sensitive sensing.