Dead Time Compensation for High-flux Depth Imaging

Abstract

Time-correlated single photon counting (TCSPC) is a powerful technique for lidar depth imaging, allowing for accurate range measurements from very low light levels. However, single-photon detectors used in TCSPC have a dead time after each photon detection, which blocks registration of subsequent photons arriving within that dead time, causing a distortion of the detection time distribution. The most common approach to avoiding dead time distortion is to optically reduce the photon arrival rate such that with high probability no photons arrive during the dead time. However, this prevents the high photon flux acquisition necessary for real-time applications such as autonomous navigation. In this paper, we propose a dead time compensation method that enables fast data acquisition with dead time-limited detectors. Specifically, we model dead time-affected detection times as a Markov chain, present a simple method for approximating the stationary distribution, and estimate depths using a log-matched filter matched to that distribution. Our method applies to multimodal imaging systems where a standard camera is used in conjunction with lidar to provide information about scene reflectivity. Simulation results for real 3D scenes show that our method reduces the root mean squared error by several orders of magnitude for the same acquisition time.