Design of a LiDAR-Camera Self-Calibration Algorithm Based on Railway Geometric Features

This paper addresses the challenges of LiDARcamera calibration in railway environments by proposing an automatic calibration method based on inherent geometric features. By utilizing railway-specific structures such as tracks and catenary poles, the method eliminates the need for artificial calibration targets. To overcome matching difficulties caused by point cloud sparsity, a two-phase strategy is adopted: initial calibration using point features followed by establishing 3D-2D geometric constraints through line feature extraction, with an iterative optimization algorithm designed to solve the optimal extrinsic matrix. The approach specifically considers practical requirements in complex railway scenarios, including dynamic operating conditions and multi-sensor compatibility.