In low-light conditions, the single-photon light detection and ranging (Lidar) technique based on time-correlated single-photon counting (TCSPC) is suited for collecting a three-dimensional (3D) profile of the target. We present a rapid 3D reconstruction approach for single-photon Lidar with low signal-to-background ratio (SBR) and few photons based on a combination of short-duration range gate selection, photon accumulation of surrounding pixels, and photon efficiency algorithm in this paper. We achieve the best noise filtering and 3D image reconstruction by choosing the optimal combined order of simple methods. Experiments were carried out to validate the various depth estimation algorithms using simulated data and single-photon avalanche diode (SPAD) array data under varying SBR. The experimental results demonstrate that our proposed method can achieve high-quality 3D reconstruction with a faster processing speed compared to the existing algorithms. The proposed technology will encourage the use of single-photon Lidar to suit practical needs such as quick and noise-tolerant 3D imaging.In low-light conditions, the single-photon light detection and ranging (Lidar) technique based on time-correlated single-photon counting (TCSPC) is suited for collecting a three-dimensional (3D) profile of the target. We present a rapid 3D reconstruction approach for single-photon Lidar with low signal-to-background ratio (SBR) and few photons based on a combination of short-duration range gate selection, photon accumulation of surrounding pixels, and photon efficiency algorithm in this paper. We achieve the best noise filtering and 3D image reconstruction by choosing the optimal combined order of simple methods. Experiments were carried out to validate the various depth estimation algorithms using simulated data and single-photon avalanche diode (SPAD) array data under varying SBR. The experimental results demonstrate that our proposed method can achieve high-quality 3D reconstruction with a faster processing speed compared to the existing algorithms. The proposed technology will encourage the use of single-photon Lidar to suit practical needs such as quick and noise-tolerant 3D imaging.

在光学三维成像领域，双目视觉三维成像、飞行时间成像、结构光等方法为深度信息获取提供了多种可能性，但上述方法均存在一定的技术局限性。偏振作为光波的特有属性，提供了更多维度的信息，将光学偏振特性应用至三维成像，可以在一定程度上弥补现有三维成像方法的不足。基于不同的三维成像原理，从偏振特性的应用方式出发，分别介绍了近年来国内外在三维成像与光学偏振相结合方面的研究思路、方法与成果，涵盖偏振双目视觉三维成像、偏振飞行时间及偏振结构光三个方面，同时对比了三类融合光学偏振的典型三维成像方法的特点、适用场景等，总结了各类方法的优缺点，为该领域的后续研究提供了指引。

对文物的保护和修复是一项覆盖考古、物理、化学及生物等多学科交叉的工作,利用各种科技手段无损地表达文物的历史意义一直是文保工作者的发展方向。太赫兹作为近二十多年来的新兴学科,被广泛应用于各个领域,例如安检、药品质量监督和半导体等,尝试将太赫兹应用到文保领域也是近年来的研究热点。主要讲述太赫兹飞行时间成像在漆器盒子样品上的应用。实验结果表明,研究的漆器盒子样品具有至少两层结构,胎体是木质材料,在木质材料的表面涂漆并在上面描绘不同的图案,达到装饰的目的。并且可以通过太赫兹飞行时间成像检测漆盒中的胎体也就是木质结构内部是否存在损坏。

深空探测器的功耗和体积有限, 任务工况多样, 与低轨道地球探测器相比, 深空探测器对导航敏感器的任务能力提出了更高的需求。提出了一种基于飞行时间成像的快速位姿测量和地物目标识别技术。为了在保证位姿测量精度的前提下满足对位姿测量时间性能的需求, 提出了一种基于深度信息的动态尺度估计方法。该方法提升了物方多尺度变化条件下点云配准的时间稳定性, 平均配准时间缩短60%以上, 平均配准精度约为0.04 m。为了满足多尺度、多形态地物目标识别的需求, 使用了基于轻量化深度神经网络, 可根据场景深度信息进行地物检测。结果表明, 该方法可对地物特征进行快速感知, 在真实场景中的准确率达到70%以上。