Source-linked AI summary
Single-photon computational 3D imaging at 45 km
Zheng-Ping Li, Xin Huang, Yuan Cao, Bin Wang, Yu-Huai Li, Weijie Jin, Chao Yu, Jun Zhang, Qiang Zhang, Cheng-Zhi Peng, Feihu Xu, Jian-Wei Pan
TL;DR
Long-range LiDAR faces weak-return and strong-noise challenges that limit imaging. The paper develops a photon-efficient super-resolution algorithm and confocal single-photon LiDAR system, demonstrating imaging up to 45 km at approximately one signal photon per pixel.
Problem
Weak return photons mixed with strong noise create challenges for extending long-range LiDAR imaging.
Method
The paper combines a confocal single-photon LiDAR system, noise suppression, and a photon-efficient super-resolution algorithm using transverse smoothness constraints.
Results
45 km active single-photon 3D imaging was demonstrated with approximately 1 photon per pixel and a signal-to-noise ratio as low as 0.03.
Takeaways & Limitations
The demonstrated approach supports long-range single-photon 3D imaging, with feasibility for extending the system to a few hundred kilometers by refining the setup.
Takeaways & Limitations
Daytime imaging is slightly blurred by increased air turbulence, and smog prevents resolving different tree layers.
Abstract
from arXiv · showhide
Long-range active imaging has a variety of applications in remote sensing and target recognition. Single-photon LiDAR (light detection and ranging) offers single-photon sensitivity and picosecond timing resolution, which is desirable for high-precision three-dimensional (3D) imaging over long distances. Despite important progress, further extending the imaging range presents enormous challenges because only weak echo photons return and are mixed with strong noise. Herein, we tackled these challenges by constructing a high-efficiency, low-noise confocal single-photon LiDAR system, and developing a long-range-tailored computational algorithm that provides high photon efficiency and super-resolution in the transverse domain. Using this technique, we experimentally demonstrated active single-photon 3D-imaging at a distance of up to 45 km in an urban environment, with a low return-signal level of $\sim$1 photon per pixel. Our system is feasible for imaging at a few hundreds of kilometers by refining the setup, and thus represents a significant milestone towards rapid, low-power, and high-resolution LiDAR over extra-long ranges.
Algorithm.
The paper addresses long-range 3D-imaging challenges from diffraction, turbulence, weak signals, and noise with a photon-efficient super-resolution algorithm. Experiments in urban environments demonstrate 45 km imaging with detailed depth reconstruction at very low photon levels.
- Challenges: The long-range setting combines diffraction and turbulence with extremely low SNR, limiting spatial resolution and signal–noise unmixing.The authors identify these as two challenges unique to long-range operation.
- Algorithm: The algorithm globally gates signal detections in time by aggregating pixel counts and searching for peaks associated with clustered scene reflectors.This filters noise before reconstruction by exploiting finite depth clustering in natural scenes.
- Algorithm: The method solves a 3D inverse problem with a modified SPIRAL-TAP solver, total-variation regularization, and a transverse-smoothness constraint.Measurements retain reflectivity and depth features across spatial and temporal dimensions.
- Results: The resulting reconstructions provide super-resolved reflectivity and depth images, including a 0.6 m transverse resolution at 45 km.Fine-interval scanning resolves small windows despite an approximately 1.0 m far-field diffraction limit.
Forward model.
The forward model represents photon-histogram measurements as a convolution of depth-reflectivity with spatial and temporal kernels, plus background noise. Reconstruction estimates intensity and depth from low-resolution data using global gating and a 3D deconvolution algorithm with transverse smoothness.
- Forward model: The scanning-angle rate function convolves the depth-reflectivity map with spatial and temporal kernels representing the field-of-view intensity distribution and laser-pulse shape.The model also includes the speed of light and background rate.
- Forward model: The photon measurements follow a Poisson model represented as a 3D matrix, with the observed data containing signal and background contributions.The discrete kernels and depth-reflectivity representation define the matrix model.
- Forward model: The reconstruction goal is to estimate the intensity-and-depth matrix from low-resolution photon-histogram measurements.The estimated matrix contains both intensity and depth information.
- Algorithm: Existing photon-efficient algorithms cannot be directly applied because their one-depth-per-pixel assumption is invalid for the long-range forward model.The paper therefore develops a computational algorithm tailored specifically to long-range 3D imaging.
- Algorithm: The proposed reconstruction globally gates signal-bearing time bins, solves a 3D deconvolution problem with modified SPIRAL-TAP, and imposes transverse smoothness using total variation.A depth map is then formed from each pixel’s average arrival time.