A fast-scanning single-photon LiDAR system for robust 3D imaging under photon-starved and interference-prone conditions

Short-dwell-time scanning and photon-starved operation make scanning single-photon light detection and ranging (LiDAR) increasingly sensitive to environmental background and system-induced artifacts. This work presents a fast-scanning, narrow-field coaxial single-photon LiDAR system that improves data spatiotemporal consistency and reconstruction robustness through scan-synchronous measurement formation. An adaptive temporal post-selection is subsequently applied to isolate effective echo windows, after which three-dimensional Poisson deconvolution with total variation regularization is performed for depth recovery. Experiments with injected background noise show that scan-induced boundary artifacts are suppressed and the ranging root-mean-square error (RMSE) is reduced by more than a factor of two relative to a conventional acquisition scheme. Outdoor tests under strong daylight at 132 m with a dwell time of only 200 μs per pixel, achieved centimeter-level depth RMSE, and additional kilometer-scale field experiments demonstrate reliable depth-resolved reconstruction under severe scattering and spatially non-uniform background interference.