Efficient and noise robust photon-counting imaging method

Authors:

Abstract
A first signal photon unit method (FSPU) was proposed in this paper which is suitable for photon-counting imaging application with low signal level and severe noise. The method exploits the different statistics of signal detections and noise detections that signal detections would cluster while noise detections tend to be equally distributed to distinguish signal from noise. For each spatial pixel, laser illumination would not stop until the first signal photon unit or the first n detections that cluster within ε range is discovered where n and ε is a preset parameter to describe what kind of cluster we wish to identify. The number of pulses to obtain this first signal photon unit is a random variable and it contains the intensity information. Depth and intensity images are reconstructed through the mean time of FSPU and maximum likelihood estimation from the number of emitted pulses respectively. Simulation verifies its feasibility even when signal to noise ratio is well below 0dB.