Automated intrinsic/extrinsic PTZ camera calibration using mobile LiDAR data

Pan-tilt-zoom (PTZ) cameras are essential for traffic management by providing dynamic surveillance. Regular PTZ camera calibration is crucial for quick and accurate visualization, but frequent maintenance and environmental factors often cause changes in their intrinsic/extrinsic characteristics. This study proposes an automated calibration approach that integrates image and mobile LiDAR data processing. The approach estimates the principal distance using homography equations. Then, a learning strategy is used to calculate an initial guess of the camera orientation and elevation using approximate planimetric coordinates and LiDAR trajectory. A constrained sample consensus approach refines the exterior orientation parameters by matching linear features between PTZ images and LiDAR data, significantly reducing the number of trials by 99%. The accuracy of the calibration is validated through improvement in the backward/forward projection by 94% and 98.5%, respectively. Additionally, pan/tilt discrepancies between actual and estimated values for pointing at a specific location were less than 1 degrees.