Shen Jiajun, Shi Yan, Zhang Yi, Xu Rui, Zhao Tianqi, Zhao Chunliu, Zhan Chunlian
Appl Opt. 2025 Apr 20;64(12):3193-3203. doi: 10.1364/AO.553230.
Light detection and ranging (LiDAR) actively senses the surrounding environment and acquires 3D information by measuring the flight time of light pulses, making it a key technology in active imaging. The single-photon avalanche diode (SPAD), known for its extremely high detection sensitivity, has been widely applied in LiDAR systems. With the advent of million-pixel time-gated SPAD array cameras, LiDAR can achieve wide-field, fast 3D imaging in low-light environments. However, compared to traditional LiDAR systems using single-pixel detectors, time-gated SPAD arrays present new challenges in 3D reconstruction accuracy due to the complexity of simultaneous multi-pixel detection and the need for high-speed data acquisition. To improve the accuracy and reliability of 3D reconstruction using time-gated SPAD array cameras, this paper theoretically analyzes the key sources of errors in the system, refines the imaging model, and proposes corresponding compensation methods. Experimental results show that after compensation, the error was reduced by more than 60%, and a depth resolution within 1 cm was achieved, verifying the effectiveness of the proposed theoretical model and calibration method and providing a reference for future research.
光探测与测距（LiDAR）通过测量光脉冲的飞行时间来主动感知周围环境并获取三维信息，使其成为主动成像中的一项关键技术。以极高的探测灵敏度著称的单光子雪崩二极管（SPAD）已在LiDAR系统中得到广泛应用。随着百万像素时间选通SPAD阵列相机的出现，LiDAR能够在低光照环境下实现宽视场、快速三维成像。然而，与使用单像素探测器的传统LiDAR系统相比，由于同时进行多像素探测的复杂性以及对高速数据采集的需求，时间选通SPAD阵列在三维重建精度方面带来了新的挑战。为提高使用时间选通SPAD阵列相机进行三维重建的精度和可靠性，本文从理论上分析了系统中的关键误差源，完善了成像模型，并提出了相应的补偿方法。实验结果表明，补偿后误差降低了60%以上，实现了1厘米以内的深度分辨率，验证了所提出理论模型和校准方法的有效性，为未来研究提供了参考。
