Review of Advances in Single-Photon LiDAR
Hanyi Zhang,Xinyu Zhao,Yicheng Zhang,Labao Zhang,Mingjie Sun
DOI: https://doi.org/10.3788/CJL202249.1910003
2022-01-01
Abstract:Significance Single-photon LiDAR is widely combined with emerging imaging technology fields such as low-light detection, ultra-long-range detection, artificial intelligence (AI ), and computational imaging, producing remarkable research results. Conventional linear detection LiDAR can be divided into different categories of scanning detection, direct detection, coherent detection and non-scanning detection. LiDAR systems based on the time-of-flight (ToF) technique are capable of ranging detection of remote targets by directly interpreting the time difference between the outbound and return laser pulses. Initially, researchers tended to achieve long ranging distance by increasing the laser output power and the aperture of transceiver system. Adopting these methods, the weight and power consumption of the LiDAR system are further increased. Besides, the pursuit of high signal-to-noise ratio (SNR) echo signals in conventional LiDAR resulted in low system frequency and weak detection ability of dynamic targets. In long-distance ranging, conventional LiDAR systems were susceptible to extreme conditions such as dust and fog weather and lost effectiveness.With the advancement of single-photon detector (SPD) and precision electronic-timing technology, LiDAR based on time-correlated single photon counting (TCSPC) appeared, which has become a new way to solve the above problems. Compared with the conventional LiDAR systems, TCSPC LiDAR systems have the characteristics of higher sensitivity, higher depth accuracy, shorter acquisition time and higher photon efficiency. By time accumulation of the single echo and photon statistics, TCSPC LiDAR does not rely on single pulse measurement results, so that it does not emphasize on high SNR of single detection pulse and high laser power. TCSPC LiDAR with the detection sensitivity reaching the single photon level is called single-photon LiDAR. In the single-photon LiDAR system, only one photon can be detected and tagged, which achieves the theoretical detection limit. This new photon statistics scheme emphasizes on the full employment of the limited echo photon information, thus improving the photon utilization rate while maintaining high sensitivity. In improving working distance and detection efficiency of the LiDAR system, TCSPC has incomparable advantages over the traditional technology.Many corresponding advances have been achieved in single-photon LiDAR systems, but they still face a series of challenges in noise suppression and performance improvement such as working range and depth accuracy. This review aims to act as an introduction to the topics of ToF-based single-photon LiDAR for the general reader and to provide a briefintroduction to the current technologies available.Progress The process of accumulating discrete echoes in single-photon LiDAR can be regarded as targets acquisition through laser pulses, which means that a few echo photons contain abundant target information. Firstly, the ToF ranging method, TCSPC technique and fundamental principle of single-photon LiDAR are summarized. Then, the single-photon LiDAR system components are introduced, including the laser, optical transceiver system, single-photon detector, TCSPC module, and the control and data processing unit. Performances of lasers applied in typical single-photon LiDAR are listed ( Table 1 ). The main abilities of single-photon LiDAR are ranging distance and detection accuracy. In order to achieve high-precision long-distance three-dimensional imaging of single-photon LiDAR, one method is to optimize the hardware part. The traditional way is to promote transmission laser power and increase the efficiency of optical transceiver system by expanding the receiving aperture and suppressing the noise of the optical transceiver structure. The main parameters of single-photon detectors include dark count rates (DCRs), dead time and detection efficiency.Image reconstruction algorithms were designed for solving the case of low echo photons in single-photon LiDAR. For example, the first-photon imaging algorithm was adopted to improve the utilization efficiency of echo photons. According to the detection probability model of echo photons in the first-photon imaging algorithm, the spatial structure and reflectivity of the three-dimensional scene are acquired, and the target information is fully reconstructed with high quality. Before imaging depth recovery, noise suppressed or removed cannot be neglected. Mainstream algorithms, such as sparse Poisson intensity reconstruction algorithm ( SPIRAL ), optimize the regularization term in different scenes to achieve the optimal reconstructed image. In the end, the main development history of single-photon LiDAR is introduced ( Fig. 8 ), and the technical applications of single-photon LiDAR in long-distance detection imaging and autonomous vehicles are discussed. At present, the farthest imaging and ranging working distance of single-photon LiDAR has been expanded from sub-kilometer level (2013) to 200 km (2021 ), and the imaging accuracy has reached millimeter level from sub-decimeter level. It should also be pointed out that the single-photon LiDAR was developed in the direction of real-time target recognition. Single-photon LiDARs are capable of dynamic targets tracking and transient images recording, which is expected to be applied in the field of autonomous vehicles and space fragment monitoring.Conclusion and Prospects Single-photon LiDAR has gained extensive attention and produced remarkable research in long-range detection imaging because of its single-photon detection and tagging ability. At present, the research on single photon LiDAR mainly demonstrates the imaging principle and reconstruction algorithm in the laboratory, and more research is needed to verify the effect and adaptability of few-photon algorithms in the actual environment. In summary, single-photon LiDAR still needs in-depth and detailed explorations to further promote the system performance and optimize new reconstruction algorithms.