Research on Estimation Method of Occlusion Target Range Parameters for Low Signal-to-Noise Ratio GM-APD LiDAR

Geiger-Mode Avalanche Photo Diode (GM-APD) LiDAR, with its single-photon sensitivity, significantly enhances the three-dimensional imaging and reconnaissance performance of targets in occlusion environments. However, in complex occlusion scenarios, laser signals are susceptible to multiple scattering, resulting in the superposition of various echoes within a single pixel field of view, which makes target range information difficult to extract. Especially under low signal-to-noise Ratio (SNR) conditions, useful signals are severely submerged by noise, further exacerbating inaccuracies in range estimation. To address these issues, this paper proposes an occluded-target range parameter estimation method based on the Neyman-Pearson criterion. The method formulates multi-peak echo detection as a binary hypothesis testing problem and constructs a likelihood ratio detection model based on the statistical characteristics of GM-APD, enabling reliable decision-making for target signal intervals. Further combining the K-means clustering and differential analysis methods, the distance values of multi-peak echoes are estimated pixel-by-pixel, thereby accurately extracting the range information of occluded targets. The simulation results show that when the signal-to-noise ratio is as low as −12 dB, the algorithm proposed in this paper improves the restoration degree by at least 12.72 % and the variance by at least one order of magnitude compared with other comparison algorithms, which can effectively improve the accuracy of the target range information under occlusion conditions.