The converted number of photoelectrons after the photocathode of scannerless imaging lidar meet the Poisson statistical distribution. The calculation for judging threshold of lidar system is proposed according to maximum likelihood estimation. Taking a typical scannerless range-gated imaging lidar as the case, an approach to approximately calculate detection probability, false-alarm probability and dropouts probability is presented which is proved by Monte-Carlo simulation. The effect of system threshold on detection probability is analyzed. A proposal that increases the detection probabilities when threshold changes with signal and noise is raised. Under the hardware limitation of lidar system, as the detection requests are satisfied, reducing the detection range and number of detector pixels mostly, increasing arrived signal photons from the target, determining the threshold by maximum likelihood estimotion, the detection probability of the system is improved.