Single-Shot Targetless LiDAR-Camera Calibration with Automatic ROI Selection and Particle Swarm Optimization

The accurate extrinsic calibration of LiDAR-camera sensor pairs is essential for reliable multi-sensor perception in autonomous systems. Manual measurements are prone to error, and extrinsic parameters can drift over time due to vibration or mechanical disturbances, consequently degrading system performance. We propose an automatic, single-shot targetless calibration framework that requires only one synchronized image-point cloud pair. The method automatically identifies edgerich regions of interest (ROIs) from the camera image, extracts geometric features in both modalities, and optimizes the 6-DOF extrinsic transformation using deterministic Particle Swarm Optimization (PSO). Experimental validation on the KITTI dataset demonstrates competitive performance compared to stateof-the-art calibration methods.