瞬态光学研究室知识库

We report a five-step processing algorithm for photon counting depth imaging under strong background noise environment, and experimentally verified that this method can realize single photon counting 3D imaging under signal-to-noise ratio (SNR) less than 1. In order to accurately locate the target position when the ambient flux is high, a computational pile-up correction is performed to recover the underlying signal photons, then performing an adaptively full-pixel target position locating. After the target position is determined, the signal and noise photons are separated pixel-wisely using a cluster method. Pixels which have no arrival photons are filled by using the information from neighborhood. At last, by using total variation spatial regularization, the depth images are reconstructed accurately. To validate the proposed method, a single photon counting 3D imaging system is established and experiments at different noise levels are carried out. Experimental results show that accurate depth imaging can be reconstructed with the SNR as low as 0.41. This approach is suitable for depth imaging under high background noise and also very suitable for the noncooperative target imaging with no prior knowledge of the target distance for its adaptive range gating.