A probability density functions convolution based analytical detection probability model for LiDAR with pulse peak discriminator

Detection probability as a key indicator of LiDAR determines the quality of 3D images. To explore the detection essence of pulse LiDAR, a probability density function (PDF) convolution based analytical detection probability model is established to predict the detection performance of pulse LiDAR with peak discriminator. In the model, the analytical detection probability is determined by convoluting the PDFs of echo pulse point and the maximum noise amplitude derived from the cumulative multiplication of the PDFs of all noise points. Experiments showed that the theoretical probabilities calculated from model is consistent with the experimental results. Based on the model, a detection threshold of peak discriminator is set to 4.5 times noise standard deviation for achieving a detection probability of 90 %@14.6 dB and a false alarm probability of 0.17 %, which is verified using the lab-built LiDAR. The presented model offers valuable guidance for system design and detection parameter selection.